Genetic knockout of porcine GGTA1 or CMAH/GGTA1 is associated with the emergence of neo‐glycans

Pig‐derived tissues could overcome the shortage of human donor organs in transplantation. However, the glycans with terminal α‐Gal and Neu5Gc, which are synthesized by enzymes, encoded by the genes GGTA1 and CMAH, are known to play a major role in immunogenicity of porcine tissue, ultimately leading to xenograft rejection.

derivatives elongated by Neu5Ac were increased in both KO groups. N-glycans capped with Neu5Gc were increased in GGTA1-KO pigs compared to WT, but were not detected in GGTA1/CMAH-KO pigs. Similarly, the ganglioside Neu5Gc-GM3 was found in WT and GGTA1-KO but not in GGTA1/CMAH-KO pigs. The applied detergent based decellularization efficiently removed GSL glycans.
Conclusion: Genetic deletion of GGTA1 or GGTA1/CMAH removes specific epitopes providing a more human-like glycosylation pattern, but at the same time changes distribution and levels of other porcine glycans that are potentially immunogenic.

INTRODUCTION
Clinical application of xenogeneic tissue holds great potential to overcome human donor organ shortage, in particular in the younger population. 1,2 Several types of porcine, bovine, and equine tissues have been used for applications in cardiac, orthopedic, esophageal, and urinary tract surgery. 2 In particular, porcine or bovine pericardium has been extensively used for the production of heart valves chemically fixed in glutaraldehyde (GA), also known as bioprosthetic heart valves (BHVs). [3][4][5][6] Unfortunately, BHVs are associated with structural valve deterioration (SVD), in particular in patients below 60 years of age. 3,6 SVD of BHVs has been correlated to GA fixation and to the presence of glycan xenoantigens. [7][8][9][10] Indeed, in addition to protein xenoantigens, carbohydrate antigens seem to play a major role in tissue immunogenicity due to immense variability, both within and between species, and considerably greater than that of proteins. 11 The immune response induced by glycan xenoantigens is mainly humoral, and driven by natural and pre-existing antibodies. 2 The major glycan structures associated with an adverse immune response are the Galα1-3 Gal (α-Gal) epitopes, found as terminal residue on N-glycans and glycolipids, and the non-human sialic acid N-glycolylneuraminic acid (Neu5Gc), also known as Hanganutziu-Deicher antigen, found in N-, O-glycans and glycolipids. 12,13 Both xenoantigens are not synthesized in humans due to inactivation of genes encoding the enzymes N-acetyllactosaminide alpha-1,3-galactosyltransferase (GGTA1) and cytidine monophosphate-N-acetylneuraminic acid hydroxylase (CMAH). Interestingly, humans express circulating IgA/IgG/IgM antibodies against α-Gal and Neu5Gc. 5 The α-Gal epitope on xenogeneic tissue has been associated with hyperacute graft rejection, whereas Neu5Gc is involved in the acute graft rejection process observed upon transplantation of xenogeneic tissue/organs in humans. 14,15 Neu5Gc has been identified in human tissues; as it can be introduced with meat-and milk-derived dietary products. Despite its incorporation in human tissues, Neu5Gc is recognized as non-self by the immune system, which in turn induces high levels of polyclonal anti-Neu5Gc antibodies. 2 Decellularization of allografts and xenografts has been successfully used to remove cellular material, with the goal to render organs suit-able for transplantation, that is, to become immunologically compatible and capable of growing within the patient. 16 Different chemical, enzymatic, and physical techniques have been tested for decellularization to remove the known immunogenic compounds, such as cells and water-soluble molecules. 16 However, for elimination of carbohydrate xenoantigens, blocking their synthesis has proven to be most efficient.
Gene-edited pigs lacking the expression of both α-Gal and Neu5Gc were produced for the construction of safer xenografts. 17,18 The use of such xenografts decreased antibody binding to porcine tissue. As an example, in vivo short-term studies using GGTA1 knockout stented GA fixed BHVs, (lacking the expression of α-Gal), transplanted into non-human primates, showed a lower immune response compared to the standard biological heart valves isolated from wildtype animals. 19 Furthermore, triple transgenic (GGTA1-KO/human CD46/human thrombomodulin) pig heart heterotrophically transplanted into baboons with various immunomodulatory strategies prevented graft rejection and resulted in graft survival beyond 900 days. 20 Other studies showed that long-term graft failure still occurs, unless using strong immunomodulatory strategies. 21,22 While genetic engineering of donor pigs can knockout the expression of known xenoantigens, it could also induce the increased expression of pre-existing sugar residues, or the de novo appearance of glycans. 23,24 Previous studies have shown that deletion of the GGTA1 and CMAH proteins resulted in an increased production of high mannose and truncated glycans, xylosylation and increased fucosylation on N-linked glycans isolated from serum proteins of GGTA1/CMAH-KO pigs. 23 These types of sugar structures are only present in low amounts in humans and might induce different types of immune responses.
In the current study, we investigated the effects of genetic knockout of two important glycan epitopes on the porcine glycome using porcine pericardium as a model. By using multiplexed capillary gel electrophoresis coupled to laser induced fluorescence detection

Tissue collection and preparation
Heart valves were either obtained from 6-months old German Landrace pigs (wildtype (WT) pigs) from a local slaughterhouse (Merhold GmbH), from GGTA1-KO pigs (Friedrich-Loeffler-Institut, Mariensee) 25 or from a GGTA1/CMAH-KO pig (School of Life Sciences Weihenstephan, TUM, Freising). 17 The parietal layer of the pericardium was isolated from the heart and the excess fat and tissue were trimmed carefully. Out of six WT porcine heart valves, three pericardia were left untreated (WT native, n = 3) and three were processed for

Decellularization
The pericardia were decellularized with a modified version of the protocol for decellularization of porcine pulmonary valves, described previously. 26

N-glycan analysis
N-glycans were analyzed by multiplexed capillary gel electrophoresis coupled to laser-induced fluorescence (xCGE-LIF) as described previously, 27

xCGE-LIF analysis combined with exoglycosidase digests enables reliable N-glycan annotation
The xCGE-LIF analysis of differentially treated pericardia (native and For several N-glycans we observed an adjacent peak migrating approximately 8 migration time units (MTU) earlier in our xCGE-LIF analyses. We could provide evidence that these peaks represented labeling artifacts caused by an incomplete labeling reaction (as described in Supplementary Data). However, as we could clearly assign the glycan belonging to these incompletely labeled artifacts, they were included in our further analyses. Altogether, it was possible to specifically annotate 19 peaks representing 11 different biantennary and core-fucosylated N-glycans with α-Gal-epitopes and/or terminal sialic acids(s) (e.g., Neu5Ac and Neu5Gc, Figure 1A). These 19 peaks covered the most intense signals in our analyses and together represented more than 47% of the total signal intensity in all 12 samples analyzed.

Comparison of relative N-glycan intensities between WT, GGTA1-KO, and GGTA1/CMAH-KO pigs revealed the emergence of neo-glycoepitopes
For the 19 peaks that could unequivocally be assigned to specific N-glycans, we compared the changes of relative signal intensities in samples of genetically-modified compared to WT pigs ( Figure 1B

Determination of normalized N-glycan intensities confirmed altered N-glycosylation between WT, GGTA1-KO, and GGTA1/CMAH-KO pigs
The entire xCGE-LIF-based N-glycan analysis was repeated upon spikein of a defined amount of the standard glycan neolactotetraose (nLc4) into each sample. Depending on its specific migration behavior, this standard glycan gives rise to a peak at approximately 466 MTU, a position where it does not interfere with sample peaks. Intensities of all peaks were then scaled to the intensity of the spike-in which was set to 1 for all samples, giving rise to normalized N-glycan intensities. Subsequently, we quantitatively determined whole N-glycan levels in native and decellularized tissues of WT, GGTA1-KO, and GGTA1/CMAH-KO animals. Therefore, we summed all normalized intensities of all Nglycan signals that were above a defined threshold of two. For WT and GGTA1-KO, N-glycan levels were considerably higher in decellularized compared to native tissues ( Figure 2). Annotation of peaks in the N-glycan analysis with spike-in (for determination of normalized intensities) was adapted from peaks that could clearly be assigned to glycan structures based on the annotation done for the N-glycan analysis without spike-in (for determination of relative signal intensities, Figure 1A) ( Figure S4A and B).
Of note, we observed considerable differences of normalized N-glycan F I G U R E 1 Analysis of relative changes of N-glycan levels caused by genetic manipulation and/or decellularization of porcine pericardia. (A) Representative xCGE-LIF fingerprints of differentially treated groups for the different genetic backgrounds as indicated. N-glycan assignment was based on automated annotation ( Figure S1) in combination with exoglycosidase digests (Figures S2-S4). (B) Relative N-glycan signal intensities for the differentially treated groups of the different genetic backgrounds as indicated. Bar graphs represent mean values and bars show the standard deviation if applicable. Biological replicates: n = 3 for WT native, n = 3 for WT decellularized, n = 2 for GGTA1-KO native, n = 2 for GGTA1-KO decellularized, n = 1 for GGTA1/CMAH-KO native, and n = 1 for GGTA1/CMAH-KO decellularized. Symbol key: blue square: N-acetylglucosamine, green circle: mannose, yellow circle: galactose, purple diamond: Neu5Ac, white diamond: Neu5Gc, red triangle: fucose, brace: position of below sugar unclear. The peaks/glycans marked with an asterisk (*) represent incomplete labeling products. intensities for annotated peaks between the different genetic groups ( Figure 4C) which essentially confirmed our findings observed for relative N-glycan levels ( Figure 1B).

Analysis of glycosphingolipid glycosylation shows efficient delipidation by decellularization
We further analyzed glycosylation of GSLs in the different samples by xCGE-LIF upon spike-in of an internal standard enabling the determination of normalized signal intensities. Based on our in-house GSL glycan database most peaks could be assigned to glycan structures ( Figure 3). It becomes obvious that normalized glycan intensities are considerably decreased in decellularized compared to native samples, irrespectively of the genetic background (peaks 4,6,7,8,13,15,16,17,20,21). For peaks 9, 11 and 22, decellularization did not affect glycan levels which can be explained by own previous observations that these glycans are a potential contamination originating from impurities of the commercial ceramide glycanase (data not shown). There are no glycan signals that could be detected exclusively in the WT samples but were lacking in samples of both mutants, which should be the case for glycans with α-Gal-epitopes. The GSL-derived glycan GM3 carrying Neu5Ac was detected in all genetic backgrounds. As expected, GM3 carrying Neu5Gc, was detected in WT and GGTA1-KO tissues but no signal was observed in the GGTA1/CMAH-KO tissues.

DISCUSSION
Porcine pericardium holds great promise to be used as a tissue matrix in xenotransplantation. However, porcine tissue is well known to be highly immunogenic in the human recipient 29 and in particular porcine pericardium has recently been demonstrated to carry high levels of immunogenic α-Gal and Neu5Gc epitopes attached to Nand O-glycans. 30 Genetic modification of the donor pig has evolved as an effective strategy to prevent synthesis of immunogenic glycans on porcine organs or tissues thereby facilitating its application in xenotransplantation. 29 Genetically modified pigs lacking the α-Gal and Neu5Gc epitopes have been successfully developed. 31 Similar to N-glycans, also GSLs from pig comprise immunogenic glycan determinants. 36 Accordingly, the α-Gal epitope has been detected on porcine GSLs derived from heart and kidney, 37 aortic and pulmonary valve cusps 38 and even pericardia. 39 In contrast to these findings, we did not detect α-Gal epitopes on porcine pericardia in the present study which might be caused by analytical limitations.
A further immunogenic sugar present in pigs and associated with xeno-rejection is the sialic acid Neu5Gc. 40,41 Neu5Gc has been detected by immunohistochemistry on porcine BHVs and in native porcine aortic valves and pericardium as well 5,39 whereas it was not found upon analysis of glycosphingolipids on porcine aortic and pulmonary valve cusps. 38 Thus, Neu5Gc has been proposed to be mainly expressed on glycoproteins. 5 However, evidence for the presence of gangliosides capped with Neu5Gc has been provided from pig hearts and kidneys 37 and pericardia. 39 Accordingly, our analytical approach revealed the presence of Neu5Gc-GM3 in pericardium of WT and GGTA1-KO animals while it was absent in GGTA1/CMAH-KO material.
In previous analytical approaches aiming at the identification of porcine xenogeneic glycan-antigens, liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) was applied for N-glycan analysis 30 while GSL antigens were structurally analyzed by LC-MSMS or proton NMR spectroscopy. [37][38][39] These methods, which are well suited for (de novo) glycan structure derivation, depend on sophisticated analytical equipment and expertise. The xCGE-LIF approach has also been widely used for N-glycan analysis 42 and we recently showed that it can be employed for glycosphingolipid profiling, 28 which required the establishment of a glycan migration time database.
Once such a database has been established, xCGE-LIF allows accurate assignment of glycan structures at medium to high-throughput at low cost. 43 It is limited by size of the database but has the advantage of rapid glycan profiling, for example, for biomedical applications.
Our glycan analyses further revealed that the process of decellularization effectively removed glycosphingolids, including Neu5Gc-GM3 from porcine pericardia. The applied detergents usually disrupt the lipid bilayer of the cells, 44 while the extracellular matrix (ECM) remains intact. 26,45,46 Accordingly, our previous studies revealed inefficient removal of carbohydrates including α-Gal-epitopes by a detergentaided decellularization process. 45,47 These findings point towards the ECM as carrier of significant glycosylation. We noted that the applied decellularization did not reduce N-glycan levels as one might expect on the first glimpse. As a potential explanation, we assumed that the decellularization breaks the cells causing removal of cytosolic components which are mostly non-glycosylated while glycosylated ECM proteins are retained. As changes in the N-glycan profile caused by deletion of GGTA1 or GGTA1/CMAH in comparison to WT were similar for native and decellularized tissues, we conclude that the decellularization does not alter the N-glycan profile.
In summary, results of the present study show that interference with the glycosylation machinery by genetic modification of the donor pig does not simply remove the targeted epitopes and thus provides a more humanized glycosylation pattern, but also leads to increased expression or even new appearance of diverse glycan structures, that is, neo-glycans. These glycans might represent neo-antigens and their immunological impact has carefully to be taken into account in xenotransplantation. The authors thank Astrid Oberbeck for excellent technical support.

Lucrezia
Open access funding enabled and organized by Projekt DEAL.

CONFLICT OF INTEREST STATEMENT
E.R. is the founder, CEO and CSO of glyXera GmbH. S.C. is employee of glyXera GmbH. glyXera provides high-performance glycoanalytical products and services and holds several patents for xCGE-LIF based glycoanalysis. The other authors declare that they have no conflict of interest.