Surface modification of pig endothelial cells with a branched heparin conjugate improves their compatibility with human blood

Corline Heparin Conjugate (CHC), a compound of multiple unfractionated heparin chains, coats cells with a glycocalyx-like layer and may inhibit (xeno)transplant-associated activation of the plasma cascade systems. Here, we investigated the use of CHC to protect WT and genetically modified (GTKO.hCD46.hTBM) pig aortic endothelial cells (PAEC) in two pig-to-human in vitro xenotransplantation settings. Model 1: incubation of untreated or hTNFα-treated PAEC with 10% human plasma induced complement C3b/c and C5b-9 deposition, cellular activation and coagulation activation in WT and GTKO.hCD46.hTBM PAEC. Coating of untreated or hTNFα-treated PAEC with CHC (100 µg/ml) protected against human plasma-induced endothelial activation and damage. Model 2: PAEC were grown on microcarrier beads, coated with CHC, and incubated with non-anticoagulated whole human blood. Genetically modified PAEC significantly prolonged clotting time of human blood (115.0 ± 16.1 min, p < 0.001) compared to WT PAEC (34.0 ± 8.2 min). Surface CHC significantly improved the human blood compatibility of PAEC, as shown by increased clotting time (WT: 84.3 ± 11.3 min, p < 0.001; GTKO.hCD46.hTBM: 146.2 ± 20.4 min, p < 0.05) and reduced platelet adhesion, complement activation, coagulation activation and inhibition of fibrinolysis. The combination of CHC coating and genetic modification provided the greatest compatibility with human blood, suggesting that pre-transplant perfusion of genetically modified porcine organs with CHC may benefit post-transplant xenograft function.

incompatibilities affecting the regulation of inflammation and coagulation 7,8 . Genetic modification of the donor pig to remove antibody targets (α1,3-galactosyltransferase gene knockout [GTKO]), regulate complement activation (expression of human regulatory proteins, CD46, CD55 or CD59), and correct molecular incompatibilities (expression of human thrombomodulin [hTBM]) has improved the survival of porcine xenografts in preclinical models 9 , but does not fully address the problem of endothelial activation and shedding of the glycocalyx. Heparin, a naturally occurring sulfated polysaccharide in the body, has potent anti-coagulant and anti-inflammatory properties and is routinely used to prevent harmful blood clotting. It produces its major anticoagulant effect by accelerating ~2000-fold the inactivation of thrombin and activated factor X by antithrombin 10 . The main limitation of heparin is that in addition to binding to antithrombin, it has a propensity to bind to positively charged plasma proteins and proteins released from platelets and endothelial cells, resulting in a variable anticoagulant response and the phenomenon of heparin resistance 11 .
Corline Heparin Conjugate (CHC) is a macromolecular conjugate composed of multiple (>20) unfractionated heparin chains. Surface-immobilized CHC significantly improves the blood compatibility of biomaterial surfaces by inhibiting coagulation and inflammation [12][13][14][15] . Importantly, this unique molecule also binds directly to the surface of living human endothelial cells under physiological conditions, without compromising their biological function 16 . However, CHC binding to pig endothelial cells and its effects on the interaction of the cells with human blood have not been examined.
In this project, we hypothesized that CHC can be used to coat pig aortic endothelial cells (PAEC) with a layer that will protect them from the pro-coagulant and pro-inflammatory environment induced by xenotransplantation. We tested this hypothesis by investigating whether coating of wild-type (WT) and GTKO.hCD46.hTBM PAEC with CHC inhibited coagulation and inflammation mediated by in vitro incubation with human plasma (model 1) or non-anticoagulated whole human blood (model 2).

Results
Human plasma induces shedding of the glycocalyx by PAEC. Untreated WT PAEC strongly expressed HS, and expression was significantly reduced by treatment with 10% human plasma for 4 hrs (p = 0.03 vs. untreated) (Fig. 1A), indicating shedding of the glycocalyx. HS expression on GTKO.hCD46.hTBM PAEC was similar to that on WT PAEC and was also reduced by human plasma treatment, although this did not reach statistical significance (p = 0.09 vs. untreated) (Fig. 1A).
In a separate experiment, WT and GTKO.hCD46.hTBM PAEC were treated with hTNFα (100 ng/ml) for 2 hrs. This treatment induced a significant loss of HS expression in WT PAEC, whereas GTKO.hCD46.hTBM PAEC showed no change (Fig. 1B), which is largely attributed to the fact that hTBM's lectin-like domain mediates protection against hTNFα 17 . Together, these results demonstrate that hTNFα at high concentration can trigger alterations of the endothelial glycocalyx in WT PAEC. hCD46.hTBM PAEC were treated either with (A) 10% human plasma for 4 hrs or (B) 100 ng/ml hTNFα for 2 hrs and stained for HS expression (green) and nuclei (blue) and visualized by confocal microscopy; mean fluorescence intensity (MFI) of anti-HS FITC was quantitated using Image J software. Incubation of WT PAEC with human plasma or hTNFα induced a significant loss of HS expression (both p < 0.05). GTKO.hCD46.hTBM PAEC incubated with human plasma for 4 hrs displayed some loss of HS (not statistically significant), whereas treatment with hTNFα showed no change in HS expression. Statistical significance was measured by unpaired t-test; data are mean ± SD of three independent experiments. Scale bar: 50 µm. (C,D) Persistence of FITC-CHC binding to PAEC after 4 hrs treatment with human plasma. (C) Untreated or (D) hTNFα-treated (100 ng/ml, 2 hrs) WT and GTKO.hCD46.hTBM PAEC were coated with FITC-labeled CHC at 37 °C for 30 min, treated with 10% human plasma for 4 hrs, and analyzed for CHC binding by confocal microscopy. Scale bar: 20 µm.
We next tested whether coating with CHC would protect PAEC from the injurious effects of human plasma. PAEC were incubated with 10% human plasma for 4 hrs and assessed for deposition of activated complement (C5b-9) and for complement-mediated endothelial cell activation (upregulation of E-selectin expression). Treatment of 'CHC-uncoated' WT and GTKO.hCD46.hTBM PAEC with human plasma significantly increased deposition of complement C5b-9 ( Fig. 2A). Human plasma treatment also induced upregulated E-selectin expression in WT and GTKO.hCD46.hTBM PAEC, although the increase was only significant in the WT cells (Fig. 2B). CHC coating significantly protected WT PAEC. Uncoated GTKO.hCD46.hTBM PAEC were significantly protected compared to uncoated WT PAEC; coating with CHC further reduced complement deposition and cellular activation, although this was not statistically significant. (D,F) CHC coating of hTNFα-activated PAEC reduces complement deposition, cellular activation and coagulation induced by human plasma. WT and GTKO.hCD46.hTBM PAEC were treated with hTNFα (100 ng/ml) for 2 hrs to induce cellular activation and shedding of the endothelial glycocalyx. After, PAEC were coated with CHC and incubated for 4 hrs with 10% human plasma, and assessed for (D) complement C5b-9 deposition and (E) endothelial E-selectin expression as well as (F) D-dimer levels in the supernatants. CHC coating significantly protected hTNFα-activated WT and GTKO.hCD46.hTBM PAEC compared to uncoated PAEC. Significance was measured by one-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001). Data are mean ± SD of three independent experiments. Supernatant samples from the assay showed significantly elevated levels of D-dimers with WT and GTKO.hCD46. hTBM PAEC (Fig. 2C). Complement deposition, cellular activation and soluble coagulation activation markers were lower with GTKO.hCD46.hTBM PAEC than with WT PAEC, indicative of reduced antibody binding and increased regulation of complement and coagulation, consistent with previous observations 17 .
Coating of WT PAEC with CHC significantly reduced human plasma-mediated deposition of C5b-9 ( Fig. 2A), E-selectin upregulation (Fig. 2B) and supernatant D-dimers concentration (Fig. 2C). Reductions in these markers were also observed when GTKO.hCD46.hTBM PAEC were coated with CHC, although the decrease was only significant for D-dimers.
We then tested whether coating with CHC would protect hTNFα pre-activated PAEC against the injurious effects induced by hTNFα and human plasma. Confluent WT and GTKO.hCD46.hTBM PAEC were treated with hTNFα (100 ng/ml) for 2 hrs and coated with CHC followed by incubation with 10% human plasma for 4 hrs. Treatment of WT and GTKO.hCD46.hTBM PAEC with hTNFα induced increased C5b-9 deposition (Fig. 2D), E-selectin upregulation (Fig. 2E) and D-dimer levels (Fig. 2F). Coating of hTNFα treated WT PAEC with CHC significantly reduced human plasma-mediated C5b-9 deposition, E-selectin upregulation and D-dimers ( Fig. 2D-F). Reductions in these markers were also observed when GTKO.hCD46.hTBM PAEC were coated with CHC. Collectively, these results indicate that CHC can protect PAEC from human plasma-mediated injury by suppressing complement and coagulation activation on the cell surface.
Surface coating of PAEC with CHC prolongs clotting of human blood. We have previously developed an in vitro model in which PAEC are grown to confluence on collagen-coated microcarrier beads and mixed with freshly drawn non-anticoagulated human blood, and the time to clotting measured. This model, which mimics the interaction between the endothelial surface of a small porcine vessel and human blood 18 , was used to determine the anticoagulant effect of coating PAEC with CHC. To confirm binding of CHC, PAEC-bearing microbeads were incubated with 50-200 µg/ml FITC-labeled CHC for 15-60 min. Confocal microscopy revealed strong binding of CHC at 100 µg/ml and 15 min incubation to the cells on microbeads, whereas binding of CHC under these conditions to control beads without PAEC was minimal (Fig. 3A).

Surface coating of PAEC with CHC inhibits activation of the human plasma cascade systems.
To study the effect of CHC coating of PAEC on activation of the human plasma cascade systems, EDTAand citrate-plasma was collected from the whole blood coagulation assays at 20 min (i.e. prior to clotting) and assayed by ELISA for soluble markers of complement activation (C5a and sC5b-9), and anti-fibrinolytic activity (plasminogen activator inhibitor-1/tissue plasminogen activator (PAI-1/tPA) complexes) in EDTA-plasma and coagulation activation (thrombin-antithrombin (TAT) complexes, D-dimers) in citrate-plasma. Baseline values of all markers were measured in EDTA-plasma from blood samples taken prior to incubation with beads. All five markers were significantly increased in plasma from WT PAEC (no CHC)/human blood mixtures (C5a, sC5b-9, PAI-1/tPA, TAT, D-dimers: all p < 0.001 vs. baseline; Fig. 4A-E). CHC coating of WT PAEC significantly reduced all markers compared to plasma from WT PAEC (no CHC)/human blood (all p < 0.001; Fig. 4A-E). Plasma from GTKO.hCD46.hTBM PAEC (no CHC)/human blood mixtures showed reduced levels of all markers compared to plasma from WT PAEC (no CHC)/human blood (C5a, sC5b-9, PAI-1/tPA TAT, D-dimers: all p < 0.001) (Fig. 4A-E). CHC coating of GTKO.hCD46.hTBM PAEC reduced levels of all markers, although this was significant only for TAT complex and D-dimers (p < 0.001 vs. no CHC, Fig. 4D,E).

Discussion
Activation of the vascular endothelium of porcine organ xenografts can be triggered by several factors, including ischemia-reperfusion, antibody binding and activation of complement. One of the most rapid changes in activated endothelium is an alteration in the configuration of the glycocalyx, with shedding of its components into the bloodstream 20, 21 . Damage to the endothelial glycocalyx causes a decrease in vascular barrier function, Nuclei were stained with DAPI. Scale bar: 100 µm. (B) Clotting time of non-anticoagulated whole human blood incubated with and without microbeads. Microbeads carrying GTKO.hCD46.hTBM PAEC significantly prolonged clotting time compared to microbeads carrying WT PAEC. Coating of WT or GTKO.hCD46.hTBM PAEC on microbeads with CHC significantly prolonged clotting time compared to uncoated PAEC. Statistical analysis was carried out by one-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001). Data are mean ± SD of four independent experiments per group. (C,D) CHC coating of microbead-borne PAEC reduces deposition of human platelets during incubation with human blood. (C) PAEC-bead samples were collected from the whole blood coagulation assays at 20 min and stained for expression of CD31 (red) and deposition of CD41-positive human platelets (green). Nuclei stained with DAPI (blue). Scale bar: 100 µm. (D) Platelet deposition was determined by counting the number of platelets on 20 randomly selected beads, and is presented as % platelet deposition relative to that observed with uncoated WT PAEC. Coating of WT and GTKO.hCD46.hTBM PAEC with CHC significantly reduced platelet deposition at 20 min and 60 min, respectively. GTKO.hCD46.hTBM PAEC showed significantly reduced platelet deposition compared to WT PAEC irrespective of CHC coating. Data are mean ± SD of at least three independent experiments. Significance was tested using one-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001).
protein extravasation and tissue damage, and increases intravascular adhesion of leukocytes and platelets to promote blood coagulation. Here, we describe the impact of two approaches to limit the pro-inflammatory and pro-coagulant activation of porcine endothelial cells by human plasma or whole blood: pharmacological, by coating the cells with the heparin conjugate CHC; and genetic, by modifying the cells to reduce antibody binding and activation of complement and coagulation (GTKO.hCD46.hTBM).
The use of heparin as a therapeutic agent has limitations due to short half-life and weak affinity, as well as its variable anticoagulant responses. There is therefore a need of heparin conjugate preparations ensuring a prolonged half-life and increased affinity. In addition, surface immobilization of such conjugates may provide the preferred biocompatibility with heparin's intact biological activity. Using two in vitro models, we showed that coating of WT PAEC with CHC markedly reduced plasma-induced cellular activation and dampened activation of the cascade systems in whole blood, resulting in a significant prolongation of clotting time. We suggest that the protective effect of CHC stems from a concentration of heparin's anticoagulant and anti-complement activity at the endothelial cell surface. It is also conceivable that the glycocalyx was at least partially preserved due to the reduction in cellular activation, and/or protected by the CHC layer, although our data do not allow definitive conclusions on this point.
Interestingly, uncoated GTKO.hCD46.hTBM PAEC were protected from plasma and whole blood to a similar degree as CHC-coated WT PAEC. This confirms the important role of these genetic modifications in prolonging hTBM PAEC further reduced TAT complex and D-dimers formation. Data are mean ± SD of four independent experiments. Significance was tested using one-way ANOVA with Bonferroni correction (*p < 0.05, **p < 0.01, ***p < 0.001).
the survival of solid organ xenografts in nonhuman primate preclinical models 9 . Nevertheless GTKO.hCD46. hTBM PAEC were not completely immune to human plasma, exhibiting some loss of heparan sulfate (albeit not statistically significant) (Fig. 1A). Importantly, CHC coating on GTKO.hCD46.hTBM PAEC augmented protection against whole blood, reducing platelet deposition and thrombin generation and further prolonging clotting time. In line with previous reports 9, 22 , our results also suggest the potential benefit of heparin-based treatment following genetically modified pig organ xenotransplantation.
In summary, our data suggest that the combination of genetically modified donor pigs and treatment of xenografts with CHC may provide optimal protection against the development of a pro-inflammatory and prothrombotic milieu in the immediate post-transplant period. The unavoidable interval between organ procurement and transplantation provides an opportunity for CHC treatment; it has been shown that ex vivo machine perfusion of porcine kidneys with CHC results in uniform distribution of CHC binding within the vasculature 15 . A remaining question is whether the CHC coating will remain for a sufficient time post-transplant to allow recovery of the endothelium. We have addressed this question using a mouse model of syngeneic kidney transplantation in which grafts are cold-stored for 5 hrs in preservation solution prior to transplant, resulting in considerable ischemia-reperfusion injury. We found that addition of CHC to the preservation solution significantly improved post-transplant graft function at 24 hrs (Nordling et al., manuscript submitted). Currently, no data exist on the long-term effects of CHC treatment and its systemic effects if shed from cells in solid organ transplantation. In addition, whether the early protective effects of CHC would improve the survival of genetically modified pig xenografts remains to be determined in preclinical xenotransplantation models.

Methods
All experiments with animals and human samples were performed in accordance with relevant guidelines and regulations. Genetically engineered pigs were generated and maintained at LMU Munich according to the German Animal Welfare Act with permission of the responsible authority (Regierung von Oberbayern; approval no. 55.2-1-54-2531-54). Human blood was drawn from healthy volunteers with informed consent and with the approval of the St Vincent's Hospital Melbourne Human Research Ethics Committee.
Endothelial cells. PAEC were isolated from WT or GTKO.hCD46.hTBM transgenic pig aorta harvested from euthanized pigs as described previously 19,23 . Porcine aortic endothelial cells were used between passages 2 and 8 in all experiments. plus supplements. Cell confluence after 5-7 days was estimated by analysis of DAPI staining using a Nikon A1R confocal microscope.
In vitro human blood clotting assay. 2 ml beads with confluent cells in a 10 ml polypropylene tube were washed with RPMI medium without supplements to remove heparin, and then incubated with starvation medium for 30 min at 37 °C. Subsequently, the PAEC/beads were either untreated or coated with FITC-labeled or unlabeled CHC in RPMI-0.5% FBS for 15 min at 37 °C. After coating, the beads were washed and incubated with freshly drawn, non-anticoagulated whole human blood at a bead-to-blood volume ratio of 1:4 (i.e. 2 ml beads and 8 ml whole blood) that closely mimics the in vivo small vessel endothelial surface-to-blood volume ratio. Human blood was drawn from healthy volunteers (two individual donors [blood groups: A-positive and O-positive] were used; single donor per experiment). The bead-blood mixture tubes were incubated at 37 °C with gentle rocking, and monitored for clotting time without any interruption. For sampling, experiment tubes were interrupted at designated time intervals (i.e. before the onset of clotting, after 20 min for WT PAEC ± CHC; after 20 min and 60 min for GTKO.hCD46.hTBM PAEC ± CHC) and 2 ml samples were withdrawn to retrieve PAEC-covered beads and EDTA-plasma.
Sampled beads were washed 3 times with PBS containing MgCl 2 and CaCl 2 and fixed for 20 min in 3.7% formalin. Following washing, the beads were stained with rat anti-porcine CD31 (MAB88371, R&D Systems) plus goat anti-rat IgG Alexa Fluor 546 (Thermo Fisher Scientific) and mouse anti-human CD41 FITC antibody. The coverage of beads with PAEC, and CD41-positive platelet deposition on PAEC, were determined by counting the number of cell nuclei (DAPI) and CD41-positive platelets on 20 randomly selected beads, on a defined rounded bead surface using a Nikon A1R confocal microscope.