Expression of Human Thrombomodulin by GalTKO.hCD46 Pigs Modulates Coagulation Cascade Activation by Endothelial Cells and during ex vivo Lung Perfusion with Human Blood

Abstract

Thrombomodulin is important for the production of activated protein C (APC), a molecule with significant regulatory roles in coagulation and inflammation. To address known molecular incompatibilities between pig thrombomodulin and human thrombin that affect the conversion of protein C into APC, GalTKO.hCD46 pigs have been genetically modified to express human thrombomodulin. The aim of this study was to evaluate the impact of transgenic human thrombomodulin (hTBM) expression on the coagulation dysregulation that is observed in association with lung xenograft injury in an established lung perfusion model, with and without additional blockade of non-physiologic interactions between pig vWF and human GPIb axis. Expression of hTBM was variable between pigs at the transcriptional and protein level. hTBM increased the activation of human protein C and inhibited thrombosis in an in vitro flow perfusion assay, confirming that the expressed protein was functional. Decreased platelet activation was observed during ex vivo perfusion of GalTKO.hCD46 lungs expressing hTBM and, in conjunction with transgenic hTBM, blockade of the platelet GPIb receptor further inhibited platelets and increased survival time. Altogether, our data indicate that expression of transgenic hTBM partially addresses coagulation pathway dysregulation associated with pig lung xenograft injury and, in combination with vWF-GP1b-directed strategies, is a promising approach to improve the outcomes of lung xenotransplantation.

Introduction

Xenotransplantation, the transplantation of tissues and organs between different species, is one promising potential strategy to address the shortage of human donor organs (1, 2). Several known xenogeneic rejection mechanisms, including antibody binding, complement activation and coagulation dysregulation have been identified as major hurdles to successful transplantation of pig organs into humans or non-human primates (3-5). Addressing these hurdles through genetic modifications to the pig, with the knock-out of the α-galactosyl-transferase gene (GalTKO) and the addition of human complement pathway regulatory proteins (CPRP, e.g. hCD46, hCD55), significantly delays rapid failure of pig lungs perfused with human blood (6, 7) and prevents hyperacute rejection of pig kidneys and hearts in non-human primates (GalTKO.hCPRP heart and kidney). However these experiments unveil coagulation cascade dysregulation, which manifests in vivo as thrombotic microangiopathy (TM) and consumptive coagulopathy (CC) (8, 9), collectively termed delayed xenograft rejection (DXR).

In recent years, mechanisms underlying DXR have been a major focus of investigation (1, 8-12). After a physiologic pro-coagulant stimulus such as injury to a blood vessel wall, natural anticoagulant mechanisms are activated that prevent excessive, non-physiologic thrombin generation beyond the area of injury. A major actor in this pathway, thrombomodulin (TBM), an endothelial surface glycoprotein, binds thrombin and facilitates thrombin-induced cleavage of the plasma Protein C (PC) into its active protease, activated protein C (APC). APC inhibits not only coagulation activation, through direct inactivation of FVa and FVIIIA, but also inflammation, apoptosis and vascular edema, acting primarily through the endothelial protein C receptor (EPCR) (13). In a xenogeneic setting, however, while pig TBM can bind human thrombin, the complex is an inefficient cofactor for activation of human protein C (14), resulting in ineffective regulation of clot propagation and hypercoagulability. In theory, expression of human TBM (hTBM) in a pig xenograft endothelium should compensate for this human-pig molecular incompatibility (10, 12, 15-17). This hypothesis motivated creation of GalTKO.hCD46 pigs designed to additionally express the hTBM ‘transgene’ at high levels in the endothelium (18-20). Encouragingly, recent orthotopic transplantation studies using life-supporting GalTKO.hCD46.hTBM pig hearts in baboons demonstrated consistent long-term survival (>180d) (20), without evidence of TM in the xenograft or CC in its recipient, confirming our prior observations in a heterotopic model (21).

In this study, we characterized hTBM expression in multiple lines of GalTKO.hCD46.hTBM pigs created using either pTBM or pICAM-2 promoter. We then asked whether endothelial expression of hTBM on the GalTKO.hCD46 modulates coagulation cascade and platelet activation in pig-to-human in vitro flow chamber and ex vivo GalTKO.hCD46 lung perfusion models, with or without blockade of the vWF-GPIb axis, which we have previously shown participates in these processes (8, 22).

Materials and Methods

Cell culture

Primary pig aortic endothelial cells (PAEC) were cultured as previously (23). Cells were used after four to six passages at 80 to 95% confluence. The endothelial cell phenotype was assessed by expression of CD31 (AbD Serotec, Biorad, Hercules, CA; #MCA1746F). PAEC from 13 GalTKO.hCD46.hTBM with hTBM under the pICAM-2 promoter, 5 GalTKO.hCD46.hTBM with hTBM under the pTBM promoter, and one GalTKO.hCD46 (“negative control”) pigs, were analyzed. Human umbilical vein endothelial cells (HUVEC, “positive control”) were from ATCC (American Type Culture Collection, Manassas, VA; PCS-100-013) and grown in Vascular Cell Basal Medium (PCS-100-030) and Endothelial Cell Growth Kit-BBE (PCS-100-040).

Experimental groups

Genetically modified pigs lacking the alpha-Gal epitope (GalTKO) and expressing human membrane cofactor protein (hCD46) were generated by Revivicor (Blacksburg, VA, USA). GalTKO.hCD46 lungs from the same genetic background (n=38, previously published (6)) served as physiologic and biochemical reference group. For the GalTKO.hCD46.hTBM constructs, hTBM was under control of either the pig ICAM-2 promoter (n=10) or the pig thrombomodulin endogenous promoter (n=3).

In additional experiments, GalTKO.hCD46 (n=6) or GalTKO.hCD46.hTBM lungs (n=8, all but one with hTBM under the ICAM-2 promoter), the perfusate was treated with αGPIb Fab (6B4, 10mg/L perfusate, provided by H Deckmyn, Kulak KU Leuven, Belgium) as in our prior report (8).

Perfusate Preparation and Lung Perfusion

Freshly collected heparinized (Heparin, 3IU/ml of perfusate) human blood was mixed 1:2 (V:V) with ABO-matched fresh frozen human plasma. Lungs were perfused side-by-side in an ex vivo lung perfusion system for up to 4 hours or until organs failed (6, 24). Lung failure criteria included pulmonary vascular resistance (PVR) increase above 600 mmHg*min/L, development of gross tracheal edema (prohibiting lung ventilation or associated with lack of oxygenation) or loss of perfusate volume (>85% of starting reservoir volume) by massive intraparenchymal sequestration as described (6, 24).

Detailed experimental methods for in vitro assays (TBM expression on PAEC and lung tissue; APC co-factor activity; thrombosis assay; ELISA) and statistical analyses are provided in the Supplemental Materials.

Results

Analysis of hTBM mRNA and protein expression levels in PAEC

Primary endothelial cells were obtained from pigs expressing hTBM transgene controlled by the pICAM-2 or pTBM promoter. Relative hTBM mRNA levels in PAEC from GalTKO.hCD46.hTBM with ICAM-2 promoter were similar or higher (0.9 to 1.6 fold, mean 1.2 fold) than those measured in HUVEC (set as 1.0). PAEC with pTBM promoter showed on average 0.9 fold (range 0.5 to 1.8 fold) the HUVEC hTBM mRNA levels; no significant differences in mRNA expression were observed between the two promoters (Figure 1A). GalTKO.hCD46 PAEC were negative for hTBM mRNA, confirming species-specificity of the assay.

Figure 1.

Relative hTBM expression and function in GalKO.hCD46.hTBM PAEC. Porcine aortic endothelial cells (PAEC) isolated from pICAM-2 (n=13) and pTBM (n=5) promoter groups were analyzed along with GalTKO.hCD46 (#504-7) PAEC and HUVEC as negative and positive controls respectively. (A) hTBM mRNA levels were measured by RT-qPCR. Values represent mean ± SEM of at least two independent experiments. (B) PAEC cell surface hTBM protein expression was analyzed by flow cytometry as shown in SI Figure 1. Results are expressed as the proportion of endothelial cells (CD31+) expressing hTBM (CD141+) compared to the isotype control. Values represent mean ± SEM of at least three independent experiments. (C) Analysis of APC generation by GalTKO.hCD46.hTBM as described in Methods. APC results were normalized to a HUVEC reference of 1.0. APC assay revealed that human protein C was effectively converted to APC in the presence of each transgenic PAEC line studied here. Values represent mean ± SEM of at least three independent experiments.

Surface expression of hTBM protein on endothelial cells was evaluated by flow cytometry (See representative flow histograms in SI Figure 1). While 32.9% of HUVEC expressed hTBM (CD141+), 93.2% of ICAM-2- (range 82.1-97.6%) and 43.1% (range 22.4 to 90.3%) of pTBM-promoter GalTKO.hCD46.hTBM PAEC exhibited human thrombomodulin protein expression (Figure 1B).

APC activity test and in-vitro xenoperfusion of GalTKO.hCD46.hTBM endothelial cells

In presence of control GalTKO.hCD46 cells, the amount of APC generated by human thrombin from human PC was minimal (0.4±0.05 the level of HUVEC set as 1.0) (Figure 1C). In contrast, GalTKO.hCD46.hTBM endothelial cells with either promoter more efficiently converted human protein C to APC compared to HUVEC (p< 0.005). APC formation efficiency generally correlated with hTBM protein expression levels by flow cytometry, with the pICAM-2 promoter group generating a significantly higher level of APC compared to the pTBM promoter group (5.6±0.6 vs. 3.6±1.1, p=0.03; Figure 1C).

In an in vitro flow chamber assay (25), GalTKO.hCD46 cells stimulated thrombosis (n=11; thrombus volume 6050±771 AU, adhesion 48±6 AU & aggregation 124±1 AU; Figure 2 A,B). In contrast, as illustrated in a representative endothelial cell line from the GalTKO.hCD46.hTBM genotype cells under pICAM-2 promoter with high hTBM expression (CD141: 93% of cells positive; MFI 2.6±1.3 vs. HUVEC) (Figure 2 C-E), thrombus volume was decreased by 84% (p=0.0001), reflecting reduction in both adhesion (by 58%) and aggregation (by 64%) (p<0.001 for both comparisons).

Figure 2.

In-vitro flow chamber perfusion of human blood over confluent porcine endothelia. (A, C) Representative images showing thrombosis formation stimulated by GalTKO.hCD46 and GalTKO.hCD46.hTBM endothelia. (B, D) Qualitatively confirmed results by 3-dimensional surface representations visually depicting thrombus volume. (E) Summary of results from replicate in-vitro xenoperfusions of porcine wild-type (WT), GalTKO.hCD46 and GalTKO.hCD46.hTBM endothelia with human blood. a.u., arbitrary units; TV, thrombus volume; Δ, relative change; SA, percent surface area coverage; FR, fluorescence ratio; WT, wild type. For Δ & P values, GalTKO.hCD46 are compared to WT, and GalTKO.hCD46.hTBM are compared to GalTKO.hCD46. Values are expressed as mean ± SEM.

Lung Graft Survival and Function

Survival of GalTKO.hCD46.hTBM lungs was similar to the GalTKO.hCD46 reference group (Figure 3). With additional αGPIb treatment, survival of GalTKO.hCD46.hTBM lungs was significantly better than GalTKO.hCD46.hTBM lungs without αGPIb (p=0.031), and GalTKO.hCD46 lungs with or without αGPIb (p=0.027 and p=0.001, respectively).

Figure 3.

Survival of porcine lungs perfused ex vivo with human blood, by experimental group. Survival was defined by hemodynamic (PVR) or physiologic (gas exchange, loss of barrier function) lung failure endpoint criteria, or attainment of an arbitrary 4 hour interval with preserved graft function. Experiments using hTBM-expressing lungs that received αGPIb-treatment all ‘survived’ to the elective termination at four hours. This survival was significantly longer than in any other group (* p=0.031 vs. GalTKO.hCD46.hTBM; p=0.027 vs. GalTKO.hCD46 + αGPIb; p=0.001 vs. GalTKO.hCD46).

Pulmonary Vascular Resistance

Relative to GalTKO.hCD46 + αGPIb lungs, hTBM-transgenic lungs exhibited wide intra-group variability in PVR between 1 and 4 hours of perfusion which was not statistically significant different, (Figure 4A) although a trend toward an earlier or higher peak rise in PVR was observed in association with hTBM lungs, with or without αGPIb-treatment.

Figure 4.

Physiologic perturbations during ex vivo lung perfusion. (A) Pulmonary vascular resistance during ex vivo lung perfusion. PVR is expressed as a function of perfusion time, by experimental group. Time 0 represents measurements obtained during the first minute of lung perfusion. Organs in all experimental groups showed a rise in PVR during the first 30min of perfusion with GalTKO.hCD46 + αGPIb lungs showing the lowest values between 1 and 4 hours of perfusion (e.g. at 1h: 140±14 vs. 229±42 mmHg*min/L, p=0.012). (B) Complement cascade activation. Plasma levels of complement activation byproduct C3a, expressed as the amount of complement fragments produced above the pre-perfusion baseline. Reference GalTKO.hCD46 +/− αGPIb experiments showed less C3a elaboration than experiments using lungs with hTBM.

Complement Activation

Complement activation in association with GalTKO.hCD46.hTBM lung perfusions both with (e.g. at 120 min: 179±31nM) and without αGPIb-treatment (165±25nM) was higher than with GalTKO.hCD46 reference lungs (48.5±11nM for GalTKO.hCD46+αGPIb, p=0.045) (Figure 4B), although complement activation was significantly delayed and attenuated in all experimental groups relative to historical GalTKO and WT lungs (typically >500nM within 30 min) (26).

Platelet Sequestration and Activation

GalTKO.hCD46 and GalTKO.hCD46.hTBM lungs without αGPIb-treatment both showed a similar immediate platelet sequestration, to about 40-45% of the initial cell counts within 5 minutes (Figure 5A). Anti-GPIb-treatment delayed the initial platelet sequestration during the first hour of perfusion when compared to non-treated lungs (e.g. at 30 min: 63±9 for GalTKO.hCD46.hTBM + αGPIb vs. 39±4 for GalTKO.hCD46.hTBM, p=0.015).

Figure 5.

Platelet counts and platelet activation during ex vivo lung perfusion. (A) Platelet sequestration, expressed as the percentage of platelets remaining in the perfusate, was similar in GalTKO.hCD46 and GalTKO.hCD46.hTBM groups during the first 2 hours of perfusion but was significantly delayed with both genetics during the first 60 minutes when αGPIb be was given (e.g. ⁰ at 30min: 60±9 vs. 40±5; p=0.036). (B) Total activation of platelets, as plasma levels of βTG, was significantly reduced in hTBM lungs when compared to GalTKO.hCD46 lungs (e.g. * at 4h: 545±125 vs. 1134±124; p=0.007). αGPIb-treatment further reduced βTG elaboration. (C) Activation of circulating platelets, as CD41+ platelets expressing CD62P, was significantly reduced in association with hTBM lungs when compared to GalTKO.hCD46 (e.g. # at 4h: 8.0± 3.6 vs. 23.2±4.6; p=0.045) and nearly completely prevented when hTBM lungs were treated with αGPIb (e.g. + at 4h: 1.7±1.6; p=0.001).

Platelet activation, as plasma beta-thromboglobulin (βTG) elaboration (Figure 5B) and platelet CD62P expression (Figure 5C), was lower in association with additional hTBM expression, an effect more prominent when hTBM-transgenic lungs were treated with additional αGPIb (e.g. Δ% CD62P+ at 240min: 1.7±1.1 for GalTKO.hCD46.hTBM + αGPIb vs. 8.0±3.0 for GalTKO.hCD46.hTBM, p=0.045).

Other parameters associated with lung xenograft failure

hTBM expression by GalTKO.hCD46 lungs was associated with higher thromboxane levels between 1 and 4 hours of perfusion (Figure 6A) with or without αGPIb-treatment. Similarly, TXB2 elaboration was significantly lower with αGPIb in perfusions using GalTKO.hCD46 versus GalTKO.hCD46.hTBM lungs (e.g. at 60min: 5.1±2.4 vs. 21.4±3.4 ng/ml, p=0.01).

Figure 6.

Physiologic perturbations during ex vivo lung perfusion. (A) Thromboxane elaboration: Plasma thromboxane B2 levels were similar for all groups at the 15min time point. Both hTBM groups showed higher TXB2 values between 1 and 4 hours of perfusion. (B) Thrombin generation: Activation of the coagulation cascade was detected by the formation of thrombin, measured in plasma as level of prothrombin fragment F1+2. No significant differences were observed between groups. (C) Blood neutrophil count: Neutrophil sequestration from the blood perfusate was not significantly different between groups. Data is expressed as change from the baseline and shown as the mean ± SEM of surviving experiments.

In association with high intra-group variability, thrombin generation (F1+2) was not significantly affected by hTBM expression (Figure 6B).

Neutrophil sequestration was not significantly altered by hTBM expression or αGPIb treatment (Figure 6C).

Wet/dry ratio

Terminal lung tissue wet-to-dry ratio (WDR) for thrombomodulin-expressing lungs was similar to reference GalTKO.hCD46 lungs (4.9±0.7 vs. 4.6±1.0; p=0.46), and was not affected by additional αGPIb-treatment (4.9±0.3 for GalTKO.hCD46.hTBM + αGPIb, 4.7±0.2 for GalTKO.hCD46 + αGPIb; p=0.63).

Analysis of transgenic hTBM expression in lung tissue

Expression of hTBM transcripts was confirmed by RT-qPCR in all evaluable samples from lungs studied by ex vivo perfusion (SI Figure 2). Although the ICAM-2 promoter was consistently associated with slightly higher hTBM mRNA expression (2.38±1.06 relative units) compared to pTBM promoter (1.5±1.5), higher levels of lung hTBM mRNA did not correlate with lung protection (prolonged survival), or decreased fibrin elaboration (F1+2) or platelet activation (βTG) or sequestration. At the protein level, expression of hTBM was confirmed in all evaluable samples using Wes analysis (SI Figure 3); no significant differences were found between the ICAM-2 (blue) and pTBM (red) promoter groups. For both pTBM (Figure 7B) and ICAM (7C, &D) promoters, moderate hTBM expression was demonstrated in the alveolar epithelium but not in small and large airways in pre-perfusion tissue. In contrast, lung endothelial hTBM expression was subjectively stronger and more ubiquitously distributed in association with the ICAM-2 promoter (Figure 7C, 7D) compared to the pTBM promoter (Figure 7B), as summarized in SI Table 1. Persistent hTBM expression was confirmed by IHC in association with both promoters in GalTKO.hCD46.hTBM lung biopsies obtained at protocol-defined intervals during ex vivo perfusion with human blood, (Figure 8B, upper panels), but not in GalTKO.hCD46 pig lung (negative control, Figure 8A lower panel); scoring for intensity of tissue hTBM expression by immunochemistry tended to show a decreasing trend over the duration of perfusion, but remained clearly detectable and prevalently expressed in terminal biopsies (Figure 8B).

Figure 7.

hTBM expression by IHC in lung tissue prior to ex vivo lung perfusion. (A) Control WT lung (747C) does not show hTBM expression. (B) shows a representative staining of a pTBM promotor lung (A111-5); (C) and (D) show hTBM stainings in lungs with ICAM-2 promoter driving hTBM expression (558-3 and 558-1). HTBM expression in the endothelium was overall strong and more ubiquitously distributed in ICAM-2 promoter lungs when compared to pTBM promoter lungs. Expression analyses are summarized in SI Table 1.

Figure 8.

Lung hTBM expression during perfusion with human blood. (A) Pig lung tissue was stained by two-color immunofluorescence for von Willebrand (green), identifying endothelium, and TBM (red) as described in Methods. Samples were randomly selected from 3 lungs in each group. Representative pictures from GalTKO.hCD46 and GalTKO.hCD46.hTBM from the ICAM-2 and TBM promoter groups at various time-points during lung perfusion with human blood. There was no expression of hTBM seen in GalTKO.hCD46 control lungs (fourth row). Lungs from both the ICAM-2 (left) and pTBM (right) promoter-derived GalTKO.hCD46.hTBM transgenic pigs showed high levels of vascular hTBM expression. (B) Slides were evaluated in a blinded fashion with respect to transgene, promoter, and time point, for the expression of hTBM on a semi-quantitative scale of 0-3. Expression of hTBM was found to decrease slightly at later time points. Each colored line represents an individual animal scored for hTBM expression. Red points indicate that hTBM was not detected in the GalTKO.hCD46 control pig lungs perfused with human blood. (C) Soluble thrombomodulin levels: While reference experiments only show a slight increase in soluble TBM levels in blood plasma, experiments using hTBM lungs show a 3-4 times higher rise, measured between 2 and 4 hours of perfusion (Final value (120-240), presumably reflecting thrombomodulin shed from the lung xenograft.

Analysis of plasma levels of APC and soluble thrombomodulin

When compared to GalTKO.hCD46 controls, no difference in APC levels was detected either by directly measuring APC enzymatic activity or by antibody capture (data not shown). Pre-perfusion levels of shTBM in human blood were consistently low and not different between groups (Figure 8C). Release of soluble thrombomodulin into the perfusate was significantly higher in GalTKO.hCD46.hTBM group than reference GalTKO.hCD46 lungs in association with the pICAM-2 promoter at final experimental time point (1.47±0.38 vs. 0.41±0.1, p<0.05), whereas a similar trend with the pTBM promoter lungs did not reach statistical significance (240’: 1.14±0.35 vs. 0.41±0.1, p=0.07) in the context of a relatively small number of observations.

Analysis of ICAM-2 vs. pTBM promoter impact on lung perfusion outcomes

In the hTBM group that did not receive αGPIB treatment, expression of hTBM was driven by either the endogenous pTBM promoter (n=3) or the ICAM-2 promoter (n=10). We asked if the promoter influenced lung outcomes by analyzing lung perfusions grouped according to the hTBM promoter. While lungs with pTBM promotor showed a median survival time (MST) of 240 min (range 49-240min, 2/3 experiments reaching elective termination; sample size too small for P calculation), hTBM lungs with ICAM-2 promotor survived with a MST of 210 min (range 17-240min, 5/10 experiments reaching elective termination; p=0.29 vs. GalTKO.hCD46 control) and control lungs of 162 min (range 5-240min) (SI Figure 4). Failure mode for the one non-surviving pTBM-promoter lung was elevated PVR; the ICAM-2 lungs failed due to trachea edema (n=2), PVR rise (n=2), or sequestration of perfusate volume in the lung (n=1). Failure mechanisms are summarized in Table 1.

Table 1.

Summary of lung failure mechanisms in thrombomodulin-expressing and reference lungs. Lungs from GalTKO.hCD46.hTBM transgenic pigs in which the perfusate was treated with αGPIb did not show any organ failure during the 4 hour experimental follow up. Data shown as the number of experiments (% of experiments).

	Experimental groups
Rejection reason	GalTKO.hCD46 (n=37)	GalTKO.hCD46.hTBM (pICAM-2) (n=10)	GalTKO.hCD46.hTBM (pTBM) (n=3)	GalTKO.hCD46 +αGPIb (n=6)	GalTKO.hCD46.hTBM +αGPIb (n=8)
Trachea edema	7 (19%)	2 (20%)	none	2 (33%)	none
Loss of perfusate	2 (5%)	1 (10%)	none	none	none
PVR elevation	15 (41%)	2 (20%)	1 (33%)	1 (17%)	none
Oxygenation failure	3 (8%)	none	none	none	none
No rejection	10 (27%)	5 (50%)	2 (67%)	3 (50%)	8 (100%)

The GalTKO.hCD46.hTBM/pTBM promoter lungs that lasted to the time point of termination exhibited a late PVR rise after 180min of perfusion that was not observed with the absence of hTBM (SI Figure 5A) (pTBM promotor vs. control at 4h: 342±62 vs. control 195±45mmHg*min/L, sample size too small for P calculation).

While hTBM lungs with pTBM promoter showed the highest C3a elaboration throughout the perfusion, only ICAM-2 lungs showed a statistically significant higher complement activation compared to control lungs at 30min of perfusion (ΔC3a: 81±23 vs. 25±4 nM, P=0.001) (SI Figure 5B).

Between 1 and 4 hours of perfusion, platelet counts with pTBM promoter lungs showed higher levels than measured in the control group (% initial platelets at 4h: 63±16 vs. 29±5, sample size too small for P calculation). At the 4h time point, ICAM-2 lungs showed a trend to lower platelet sequestration when compared to the control (% initial platelets: 52±14 vs. 29±5, P=0.06) (SI Figure 6A).

While βTG-levels were similar during the first 30min in the ICAM-2 and control groups (SI Figure 6B), βTG elaboration was reduced with ICAM-2 lungs at later time points, reaching statistical significance at 2 and 4 hours of perfusion (ΔβTG at 4h: 666±104 vs. 1134±124 IU/mL, p=0.03). The lowest βTG plasma levels were found with pTBM promotor lungs (ΔβTG at 4h: 384±274 IU/mL).

Measurement of by CD62P expression on circulating platelets mirrored the result found with βTG analysis but did not reach statistical significance due to high intra-group variability (SI Figure 6C; Δ%CD62P for ICAM-2 vs. control at 4h: 9.7±5.7 vs. 20.1±3.8, p=0.18).

Thrombin formation (ΔF1+2) among hTBM-expressing lungs with ICAM-2 promoter were statistically similar to those of the control group throughout the time of observation. hTBM lungs with pTBM promoter showed high ΔF1+2 values at 4 hours (112.2±79 vs. 21.96±4.9nM for controls) (SI Figure 7A) which can be attributed to one outlier with exceptionally greater thrombin formation (209nM).

Plasma thromboxane B2 levels were elevated in both hTBM groups compared to the reference group in the first hour of perfusion (at 1h: ICAM-2 group 28125±5728 vs. control 12223±2113ng/mL; p=0.003 and pTBM group 40457±1912; n too small for statistical analysis) (SI Figure 7B).

Neutrophil sequestration did not statistically differ between hTBM promotor groups and GalTKO.hCD46 control group (SI Figure 7C).

Wet/dry ratio did not show any difference between ICAM-2 and pTBM promoter lungs (4.9±0.9 for ICAM-2 vs. 5.0±0.4 for pTBM).

Discussion

In allotransplantation and even more prominently in xenotransplantation, antibody binding, complement activation, inflammation and ischemia/reperfusion injury can each trigger or amplify activation of the coagulation cascade within blood vessels where clotting is normally highly regulated by potent anticoagulant and thromboregulatory mechanisms. Thrombin is generated during coagulation cascade activation, which promotes fibrin formation and further amplifies the coagulation processes. Physiologically, a thromboregulatory pathway is triggered by the enzymatic cleavage of protein C into APC by thrombin, teleologically preventing physiologically inappropriate thrombin generation through inactivation of coagulation factors Va and VIIIa by APC, and thereby acting as a brake on further thrombin formation and thus inhibiting intravascular clot propagation. Essential components of efficient thromboregulation involve thrombin interacting with TBM, the endothelial cell protein C receptor (EPCR), protein C, APC and protein S. In humans, thrombin binds to hTBM; the thrombin/TBM complex then activates protein C approximately 1000 times faster than does free thrombin (27). In a xenogeneic setting, however, because pig TBM is a poor cofactor for activation of human protein C, the formation of APC is deficient.

Previous studies by us and others indicated that prolific activation of the coagulation system occurs after pig organ transplantation into a non-human primate, and during perfusion of a pig organ with human blood (6, 28, 29). Moreover, in vitro experiments comparing cells that express human or pig TBM demonstrated that pig TBM does not function efficiently with human thrombin (12, 14). These findings led to the hypothesis that overexpression of human TBM in a pig xenograft could address the molecular incompatibility of the protein C pathway between species and help prevent microvascular thrombosis in xenotransplantation. Recent orthotopic transplantation studies using life-supporting GalTKO.hCD46.hTBM pig hearts in baboons demonstrated consistent long-term survival (>180d) (20), without evidence of coagulation pathway dysregulation in the xenograft or its recipient, suggesting a beneficial role for transgenic hTBM in xenotransplantation (30). However, the effect of TBM in the lung on coagulation parameters during perfusion with human blood has not previously been investigated.

Here, we evaluated whether expression of human TBM transgene by pig cells and organs is able to restore physiologic APC formation and attenuate coagulation dysregulation. Our results confirm prior reports that endothelial cells expressing hTBM support activation of human protein C by human thrombin in vitro and show suppression of clot propagation in a microfluidic flow chamber model of ex vivo perfusion. These findings are consistent with the well-established anti-coagulant role of APC, via inactivation of coagulation factors Va and VIIIa, leading to down-regulation of additional thrombin generation and inhibition of downstream fibrin formation (31). The anti-coagulant role of hTBM was further confirmed in our ex vivo lung xenoperfusions in that platelet activation markers (plasma βTG, platelet surface p-selectin) were decreased during perfusion of lungs expressing hTBM. Importantly, additional blockade of the platelet GPIb receptor in conjunction with transgenic hTBM was necessary to unveil an additive protective effect, manifest as an increase in lung survival time.

Our observations regarding hTBM effects in xenotransplantation build upon prior concordant findings reported by others. Miwa, et al (32) and Kopp, et al (33) demonstrated that wild-type PAECs transfected with human thrombomodulin significantly suppressed prothrombinase activity and delayed human plasma clotting. Iwase et al (34) demonstrated that platelet aggregation induced by GalTKO.hCD46.hTBM PAEC was significantly less than that induced by wild-type or GalTKO PAECs. In a pig-to-baboon heterotopic heart transplant model, Iwase, Ekser and Satayananda (35) and Mohiuddin et al (21) demonstrated prevention of thrombocytopenia and decreased fibrinogen in heart xenografts that expressed hTBM; in Mohiuddin’s report prolonged graft survival (298 day median survival) was accompanied by absence of thrombotic microangiopathy. In contrast, in the absence of hTBM expression heart xenografts failed at about 70 days in association with intra-graft microthrombi (30). Altogether, these observations suggest that expression of hTBM is beneficial to prevent coagulation dysregulation and protect pig organs from xenogeneic injury.

In our experiments, hTBM expression was unexpectedly associated with elevated production of several markers of inflammation and cell activation during lung perfusion, including F1+2, C3a, and TBX. F1+2 is a prothrombin peptide fragment released by cleavage of prothrombin during thrombin formation, and thus directly reflects thrombin generation. However thrombin elaboration, which initially occurs upstream from TBM and requires factor Va (among others), is potently inhibited by thrombin complexed to TMB, an effect amplified by APC, which is in turn catalyzed by TBM interacting with EPCR; inefficient interaction of pEPCR with hTBM at the level of the hTBM/thrombin/pEPCR interaction with human protein C may have blunted APC production and downstream APC-mediated inactivation of factor Va, resulting in failure of expression of hTBM by the lung to blunt prothrombin amplification and accelerated fibrin formation, and the increased F1+2 levels observed here. (As noted in Results, it is also possible that one outlier experiment skewed the F1+2 statistics, and that this apparent effect is less impactful.) Similarly, a decrease in C3a levels might have been expected because TBM directly inhibits complement activation and reduces complement-mediated cytotoxicity, one example of anti-inflammatory crosstalk between the complement and coagulation cascades (36, 37). C3a levels were quite low in all groups studied here relative to historical WT and GalTKO reference groups (6, 26), but they were not lowered further -- and indeed tended to be increased -- in association with TBM expression, suggesting that hTBM is not sufficient to inhibit complement during lung xenoperfusion.

The unexpected higher pulmonary vascular resistance (PVR) observed in association with hTBM expression correlated with increased TXB2 production, implicating increased elaboration of proinflammatory eicosanoids by activated macrophages and/or platelets, as shown previously (38). Thrombin-activatable fibrinolysis inhibitor (TAFI) is a carboxypeptidase which cleaves basic C-terminal amino acid residues of its several substrates, including fibrin (39), C3a, C5a, osteopontin, and bradykinin (40, 41). When bound to thrombin, TBM increases activation of TAFI (14); by removing plasmin’s binding site on fibrin, activated TAFI prevents plasminogen-mediated fibrinolysis and thus may paradoxically amplify local clot propagation. Accelerated TAFI activation by hTBM may also have accelerated degradation of bradykinin, an endothelium-dependent vasodilatory peptide (41), as an additional possible mechanism to explain the relative increase in PVR associated with hTBM. Our findings show that, in our ex vivo lung model and at least for these gene constructs, any anti-inflammatory, anticoagulant effects that we predicted might be associated with hTBM expression are outweighed by enhanced proinflammatory (C3a, TBX) and pro-coagulant effects (F1+2).

Endogenous thrombomodulin is shed from the endothelium under inflammatory conditions (17). Detection of soluble human thrombomodulin (shTBM) during the ex vivo perfusion of hTBM-expressing lungs presumably demonstrates ‘physiologic’ shedding from the injured transgenic endothelium. However, prevalent expression of hTBM was still readily detectable on lung endothelium by immunochemistry at the end of the perfusion, suggesting that loss of hTBM expression does not fully account for the observed increase in several inflammatory markers.

In presence of functional hTBM, protein C activation is further enhanced approximately 20-fold by binding to hEPCR (27, 42). Whether endogenous pig EPCR is functionally compatible with hTBM expressed by transgenic pig endothelial cells was recently examined using COS cells transfected with hTBM and either pig or human EPCR (43). The authors found that endogenous pEPCR is generally physiologically compatible with the human TBM/thrombin complex. Our failure to detect an increase in APC levels after xenoperfusions suggests that either APC production was not efficiently augmented under our experimental conditions in association with hTBM, or that APC was avidly bound by pEPCR. Although in theory heparin may blunt APC elaboration by interfering with function of the TBM/thrombin complex, we observed potent inhibition of thrombus formation even in the presence of high heparin concentrations in in vitro flow chamber experiments (18 IU/ ml, vs. 3 IU/ml ex vivo ), arguing against heparin suppressing APC elaboration as major deleterious factor in our models. Our unpublished data indicated a strong cytoprotective effect for PAEC’s expressing transgenic hECPR, even in absence of hTBM (Donald Harris, in preparation). We conclude that further investigations with organs and endothelial cells expressing transgenic hTBM, hEPCR and both molecules are warranted to determine whether co-expression of both molecules is necessary and sufficient to fully leverage the potentially protective role of the APC pathway in xenotransplantation of the lung or other organs. Based on our observations to date we predict that co-expression of hTBM and hEPCR in the lung may synergistically leverage the protective features associated with coordinated hTBM- and hEPCR-mediated thromboregulation and APC-mediated cytoprotection.

Lung survival time during ex vivo perfusion was only prolonged when the platelet GPIb receptor was inhibited to prevent its interaction with vWF. We previously showed that the interaction between human platelets and pig vWF are important factors during lung (8) and liver (22) ex vivo perfusions. Our similar data in presence of hTBM indicate that the GPIb-vWF axis plays a non-redundant role in promoting lung injury in the context of hTBM as evaluated here, and demonstrate that addressing non-physiologic adhesion and activation of human platelets by this pathway was required to unveil efficacy to hTBM to protect GalTKO.hCD46 endothelium from other aspects of dysregulated coagulation. Further studies using pigs expressing both hTBM and hEPCR at consistently high levels are warranted to determine the contribution of GPIb-vWF interactions in the context of optimized coagulation regulation.

In the present study, we evaluated a large cohort of transgenic pigs in which hTBM expression was driven by the endothelial cell-specific intercellular adhesion molecule-2 (ICAM-2) gene promoter, and a smaller cohort with the endogenous pig TBM promoter (16). As expected, the TBM transgene was expressed on cultured endothelial cells. hTBM was strongly expressed at the transcriptional and protein level with the ICAM-2 promoter, in line with the enhancer activity associated with this promoter’s intronic P(8) site of the ICAM-2 promoter (44). Differences in TBM expression might be expected with different promoters, and could consequently be associated with an influence on physiologic or biochemical outcomes. Although hTBM mRNA expression tended to be higher in lung tissue from the pICAM-2 group, protein expression in lung tissue was not significantly different between pICAM-2 and pTBM groups. Moreover, promoters were not associated with significant differences in physiologic ex vivo outcomes, at least in absence of GPIb blockade. Further, increased hTBM expression levels across the study did not directly correlate with longer lung survival or reduced coagulation pathway activation. All together, our findings do not suggest any advantages for (pig) ICAM-2 or the endogenous pig TBM promoter. Rather, presence of transgenic human TBM on the donor graft endothelium in the range present in the PAECs studied here was associated with physiologically meaningful hTBM activity, and both promoters were associated with similar physiologic effects. As a limitation to our study, it is possible that a hypothetical advantage in regard to perfusion outcomes that might be associated with the pTBM promotor phenotype may be masked by the relatively small number of pigs with this promotor that were available to us. Unfortunately, our data so far do not provide clear evidence to support this hypothesis.

Our studies identified significant variability between pigs in hTBM expression levels for both promoters in cultured aortic endothelial cells; and in lung tissue for the ICAM-2 promoter. Since a pool of hTBM cells, as opposed to a single cell clone, was used to generate the pigs used for this study, mRNA expression levels were not homogeneous within each particular promoter construct, as might be expected for genetically identical clones (SI Figure 3). Instead, among pigs with the same promoter, differences in random gene integration sites and copy number are likely primarily responsible for the differences in TBM expression level observed in the pigs evaluated here. Despite evidence that hTBM mRNA levels were significantly higher than in human lungs (up to 5-fold in some lungs), hTBM protein levels (corrected to actin) remained 5 to 10-fold lower. Thus it is possible that the levels of hTBM expression in the lungs studied here was insufficient to fully express the expected protective phenotype with respect to thromboregulation and inhibition of inflammation. Newer constructs with CRISPR-targeted integration sites should yield more predictable and perhaps higher hTBM expression under one or both of candidate promoters tested here, and allow us address this hypothesis in future work.

In summary, the present study confirmed functional activation of human protein C by pig endothelial cells obtained from multiple different lines of GalTKO.hCD46.hTBM pigs. Although expression of hTBM at the protein level was relatively variable between pigs, transgenic hTBM was uniformly functional in vitro across the range of protein expression levels observed in these pigs, as measured by APC production and inhibition of thrombosis; and was consistently associated with decreased platelet activation ex vivo . Although hTBM expression in the lung was associated with some potentially adverse biochemical effects, including elevated thromboxane and complement elaboration and increased coagulation activation during ex vivo pig lung perfusion, that may have influenced physiologic behavior (increased PVR), increased survival was observed for hTBM-expressing lungs when additionally targeting the GPIb-VWF axis. Overall, our data indicate that expression of transgenic hTBM, in combination with other mechanism-directed protective strategies, remains a promising approach to improve the outcomes of lung xenotransplantation.

List of non-standardized abbreviations:

APC

Activated protein C

βTG

β-thromboglobulin

CC

Consumptive coagulopathy

DXR

Delayed xenograft rejection

ELISA

Enzyme-linked immunosorbent assay

F1+2

Prothrombin fragments 1 + 2

GalTKO

α1,3-galactosyl transferase knockout

hCD46

Human membrane cofactor protein, hMCP

hTBM

Human thrombomodulin (CD141)

pTBM

Pig thrombomodulin

HUVEC

Human umbilical vein endothelial cells

ICAM-2

Intercellular adhesion molecule 2

MST

Median survival time

OCT

Optimal cutting temperature

PAEC

Pig aortic endothelial cells

PC

Protein C

PVR

Pulmonary vascular resistance

TM

Thrombotic microangiopathy

WDR

Wet/Dry Ratio