Xenoreactive antibodies in α‐granules of human platelets bind pig liver endothelial cells

Pig liver xenotransplantation is limited by a thrombocytopenic coagulopathy that occurs immediately following graft reperfusion. In vitro and ex vivo studies from our lab suggested that the thrombocytopenia may be the result of a species incompatibility in platelet glycosylation. Realization that platelet α‐granules contain antibodies caused us to reevaluate whether the thrombocytopenia in liver xenotransplantation could occur because IgM and IgG from inside platelet α‐granules bound to pig liver sinusoidal endothelial cells (LSECs). Our in vitro analysis of IgM and IgG from inside α‐granules showed that platelets do carry xenoreactive antibodies that can bind to known xenoantigens. This study suggests that thrombocytopenia occurring following liver xenotransplantation could occur because of xenoreactive antibodies tethering human platelets to the pig LSEC enabling the platelet to be phagocytosed. These results suggest genetic engineering strategies aimed at reducing xenoantigens on the surface of pig LSEC will be effective in eliminating the thrombocytopenia that limits survival in liver xenotransplantation.


INTRODUCTION
Liver transplantation using pig livers could eliminate the shortage of donor livers and allow all patients who need this form of therapy to receive one in a timely fashion.2][3][4] One potential approach to genetic engineering is to eliminate an inciting event in xenograft failure, and it is more critical than ever to understand the mechanisms responsible for triggering the initial events so that appropriate genetic engineering strategies can be proposed to produce pigs with the modifications necessary to achieve a successful patient outcome.
[7] The mechanism responsible for the thrombocytopenia is incompletely understood.Initial investigation into the thrombocytopenia using ex vivo perfusion and in vitro studies determined that the sinusoidal endothelium of the pig liver xenograft phagocytosed human platelets. 8,9The initial study used washed platelets and there was no antibody in the supernatant that the platelets were suspended in, and there was no detectable IgM or IgG on the platelet surface suggesting that antibody binding did not play a role in platelet phagocytosis by the pig liver sinusoidal endothelial cells (LSECs).Evaluation of pig livers with two xenoantigen deletions, the α-gal epitope and Neu5Gc were evaluated for platelet phagocytosis and perfusion of human platelets through GGTA1/CMAH knock out (KO) pig livers resulted in significantly less platelet uptake than wild type livers. 10We evaluated whether the asialoglycoprotein receptor (ASGR) could play a role in the phagocytosis that occurs with human platelets and pig LSEC using in vitro studies first, and then later utilizing ASGR1 KO pig livers perfused with human platelets. 11,12ASGR1 KO pig livers phagocytosed fewer human platelets than wild-type livers, but more than GGTA1/CMAH KO pig livers. 10,13[16][17] Deleting a protein with a central function in the liver should be a genetic engineering strategy of last resort.A more specific genetic strategy that gets at the exact mechanism of phagocytosis without interrupting liver function would be a much more attractive option for clinical xenotransplantation.
The use of infused human clotting factors to overcome the thrombocytopenia in preclinical pig-to-baboon models and achieved occasional improvements in survival to 29 days in one case. 18,19Death in the longer surviving recipients occurs usually as the result of thrombosis of a major vessel. 18,20The implication is that the cause of death is related to the systemic pharmacologic therapy used to overcome the thrombocytopenia.
Since we have been unable to determine a metabolic cause for thrombocytopenia in liver xenotransplantation, and platelets do have anti-pig antibodies in their α-granules, we reevaluated whether platelets might be the source of donor specific antibodies that could tether the human platelet to the LSEC in this study.We evaluated the specificity of antibodies to determine whether there was anti-pig antibody present and evaluated platelet binding to wild type and triple knockout LSEC to determine whether removal of xenoantigens could reduce platelet binding to LSEC.The results from these experiments indicate that human platelets have IgM and IgG in their α-granules that could serve to tether the platelet to the LSEC, and as such deletion of xenoantigens in the donor pig should serve to eliminate the thrombocytopenia.

Platelet preparation
Approximately 4 × 10 11 human platelets, less than 6 days from isolation and stored at 20 to 24 • C with gentle agitation, were purchased as platelet-rich plasma from the University of Miami Blood Bank, Miami, Florida, USA.For in vitro assays, approximately 1 × 10 9 platelets were washed three times in sterile phosphate-buffered saline (PBS) containing anticoagulant citrate dextrose (centrifugation at 5000 g for 5 min).Platelets were labeled with carboxyfluorescein succinimidyl ester (CFSE; Invitrogen) according to the manufacturer's protocol and then counted using a Bright-Line hemocytometer (Hausser Scientific).

Immortalized porcine LSEC
Porcine LSEC were isolated and immortalized as described previously. 21

In vitro platelet binding assay
Immortalized wild-type (WT) or TKO LSEC were cultured at 37

Microscopy
Cells and platelets were evaluated by light microscopy using an

Isolation of platelet antibody
Platelets were washed three times with sterile PBS.Platelets were resuspended and activated for 20 min at 37C with thrombin (5 IU/mL; Sigma Aldrich) to initiate degranulation and release of antibody into the supernatant.Samples were centrifuged at 17 000 g to collect platelets.Cleared supernatants were used for analysis of released antibodies by cross matching, western blotting and glycan array analysis.

Platelet uptake of gold labeled antibody
Platelets were washed three times with sterile PBS.Platelets were resuspended in PBS containing goat anti-human IgG antibody conjugated with 10 nm gold beads (Electron Microscopy Sciences) and incubated for 45 min at 37C.Platelets were washed three times with sterile PBS prior to fixation and preparation for electron microscopy (as described above).

Western blotting
The Auto Western Testing Service was provided by RayBiotech, Inc.
(Peachtree Corners, Georgia, USA) samples were resolved by size in a separation matrix, immobilized, and then probed with anti-human IgG or IgM antibody.An HRP-conjugated secondary, followed by chemiluminescent substrate incubation, allowed for detection of the target protein by electropherogram and creation of a digitized WB image.

Glycan microarray
Microarray screening studies were performed using a RayBio Glycan Array 300 Kit (RayBiotech, Peachtree Corner, Georgia, USA), which consists of two blocks of 300 synthetic glycan spots, along with relevant control spots, each 3-fold replicated.The

Platelet α-granules contain IgM and IgG that bind to three known xenoantigens
Flow cytometric analysis of human platelets detected CD62p, IgM and IgG, indicating the presence of α-granules, IgM and IgG (Figure 1A).
Confocal analysis of platelets confirmed that IgM and IgG was colocalized in CD62 containing granules (Figure 1B).TEM revealed that platelets could take up gold labeled antibody in the supernatant and package them inside their α-granules (Figure 1C).
IgM and IgG released from α-granules in human platelets was confirmed using Western blotting (Figure 2A).Isolated IgG and IgM were applied to glycan arrays to determine whether the antibodies bound to xenoantigens.Array data showed that antibodies released from the α-granules had IgM and IgG specific for three key xenoantigens (α-gal, Neu5Gc, and Sd a ) (Figure 2B).

Fewer human platelets bind to TKO pig LSEC devoid of three xenoantigens (α-gal, Neu5Gc, and Sda) than wild-type pig LSEC
We used LSEC with three xenoantigens deleted (α-gal, Neu5Gc, and Sd a ) (Figure 2C) to show a reduction in platelet antibody reactivity to the TKO LSEC as compared to the wild type LSEC (Figure 2D).Confocal evaluation of CFSE labeled human platelets showed significantly less binding to the TKO cells than to wild type (Figure 3).Confocal microscopic analysis demonstrated that platelet interaction with LSEC was associated with IgM and IgG (Figure 4A).

DISCUSSION
[24][25] The unanswered question from perfusion studies is the origin of the antibody in these experiments.It turns out that human platelets have IgG in their α-granules. 26,27The antibodies in the α-granule might have served to tether the platelet to the endothelium by binding to the platelet Fc receptor (Figure 4B).Our data confirms that αgranules in platelets have IgM and IgG.Our microscopy data generally showed that IgG labeling colocalized with the marker for α-granules, however, within all replicates the IgM labeling partially colocalized with α-granules.While we did not explore other compartments within platelets it is possible that the brightness of the label created the LSEC were grown in LSEC culture medium (RPMI medium supplemented with 10% fetal bovine serum (v/v), 100 μg/mL endothelial cell-specific growth factor, penicillin, and streptomycin) and plated onto attachment factor (Invitrogen)-treated 10 cm culture dishes at a density of 2 million cells per plate.The triple KO (TKO) LSEC were created by targeting the GGTA1, CMAH, and β4GalNt2 genes in immortalized LSEC with CRISPR/Cas9 guide RNA as described previously (Estrada et al.).
manufacturers protocol was followed to perform the antibody-based detection of IgG or IgM captured on the array spots.Antibody was diluted to 0.1 mg/mL in PBS, and then incubated overnight at 4 • C to allow binding to the array block.Detection of antibody bound spots was enabled using anti-human IgG alexafluor 555 (A21433, Thermo Fisher) or IgM alexafluor 647 (109-605-190, Jackson ImmunoResearch) at 1:500 dilution.Detection (imaging) of fluorescing spots was performed with a GenePix 4000B scanner with PMT of 750 and resolution of 5. Median fluorescent signal intensity, after local background subtraction, was normalized according to the manufacturers recommendation and used in t-tests to determine binding significance of human IgG or IgM antibody.

F I G U R E 1
Human platelets contain Cd62p positive alpha granules that contain human IgG and IgM.(A) Flow cytometric analysis of antibody containing alpha granules (CD62p) in human platelets.(B) Confocal analysis of IgG or IgM indicates that antibody colocalized with CD62p positive alpha granules inside human platelets (dashed line indicates platelet membrane)(representative images taken at 100× magnification with 4× digital zoom).(C) Transmission electron microscopy of platelets that have taken up antibody into granules.Yellow arrows indicate 10 nm gold bead labeled antibody.

F I G U R E 2 F I G U R E 4
Human platelet antibodies bind to porcine LSEC.(A) Western blot analysis of intact human platelets and antibody released from platelets, specific for human IgG or human IgM.Black arrows indicate immunoglobulin heavy chain.(B) Array analysis of antibody released from platelets indicates specificity for xenoantigens αGal, neu5Gc, and Sda (relative fluorescence units normalized to positive control).(C) Phenotypic analysis of WT and TKO porcine LSEC for endothelial cell marker, CD31, and xenoantigens aGal (IB4 lectin), Neu5Gc (anti-Neu5Gc antibody), and Sda (DBA lectin).Filled grey peaks are unlabeled cells for IB4 and DBA lectins and isotype control antibody for anti-CD31 and anti-neu5Gc antibody.(D) Flow cytometric crossmatch analysis of released platelet antibody against wild type or TKO porcine LSEC.Filled grey peaks are TKO LSEC incubated with only the secondary antibody.LSEC, liver sinusoidal endothelial cells; TKO, triple KO; WT, wild-type.F I G U R E 3 Porcine TKO LSEC bind fewer human platelets than WT LSEC.Human platelets labeled with CFSE (green) were incubated with wild type LSEC or TKO LSEC.(A) Fluorescent microscopy was used to image five fields (10×) for each slide of three experiments.(B) Fluorescent platelets bound to LSEC (∼80% confluent) and release the CFSE dye in the local area of binding resulting in WT LSEC bearing more green fluorescence than TKO LSEC.LSEC, liver sinusoidal endothelial cells; TKO, triple KO; WT, wild-type.Platelet interaction with LSEC is associated with antibody release from platelet granules.(A) Left panel: Human platelets (green, CFSE) bound to WT LSEC (red, anti-CD31 antibody).Center panel: IgG or IgM (cyan, anti-IgG or IgM antibody) in alpha granules or released from the platelet bound to the cell surface.Yellow arrows indicate the colocalization of an intact platelet with antibody containing granules and white arrows indicate antibody released and bound to the cell surface outside of human platelets.Right panel: shows the location of the fluorescently labeled bound platelet alone.(B) Illustration of human platelets when containing human IgG and IgM in granules in circulation when not activated (First panel).Activated human platelets interacting transiently with antibody to induce release of granules and antibody which bind to the cell surface.The porcine LSEC internalizes the human platelet into a phagosome for degradation.LSEC, liver sinusoidal endothelial cells; WT, wild-type.appearance of localization beyond the α-granule or the marker for the α-granule was in an upper or lower layer of the focal plane.It is also not clear if the CD62p marker is present on all α-granules.The repertoire of IgM and IgG inside the α-granules contains antibodies specific for the three known xenoantigens (α-gal, Neu5Gc, and Sd a ), and thereis a measurable decrease in the binding of the intra-platelet antibody to TKO LSEC as compared to wild type LSEC.Taken together with the decreased platelet binding to TKO LSEC compared to wildtype, our results support the concept that anti-pig antibody from inside the platelet was released and bound to the LSEC subsequently tethering the platelet to the LSEC, possibly by binding to the IgG Fc receptor on the surface of the platelet.Our results combined with our previous ex vivo liver perfusions were performed without drugs used to prevent rejection.In in vivo pig-to-NHP experiments the use of cobra venom factor, anti-CD154 could be additional factors contributing to thrombocytopenia.9,18Understanding the mechanism responsible for the binding the of the pig LSEC and human platelet is important because it will direct the genetic engineering strategy to eliminate this barrier to liver xenotransplantation.Our data suggests that the xenoantigen deletion strategy will eliminate the phagocytosis of platelets, overcoming the most proximal barrier to clinical translation of liver xenotransplantation.In the in vivo scenario, most of the antibodies that tether the platelet to the pig LSEC will come from the blood and not the platelet, but in the scenario where pretransplant antibody absorption is performed, there will still be significant ability of the human platelet to bind to the pig LSEC because of anti-pig antibodies in the α-granules of the platelets.Experience in the pig-to-NHP supports the role of antipig antibodies in α-granules in settings where antibodies are removed from the circulation.Technical limitations of the assays we used made it difficult for us to show antibodies from inside the platelet binding to pig LSEC and the human platelet.Next steps for evaluating the anti-pig antibodies in the α-granules of platelets for this should consider perfusion studies using TKO and wild type pig livers with human platelets.If these studies confirm our in vitro findings, they will lead in two directions, (a) transplantation in the pig-to-NHP model, and (b) transplant of TKO pig livers into decedents.The NHP model can be used to evaluate the longer-term impact of eliminating thrombocytopenia and the need for infusing pharmacologic doses of clotting factors.The decedent model, if successful for several days could pave the way for a clinical trial using TKO pig liver xenografts to bridge patients with decompensated chronic liver failure long enough to receive a liver allograft.