Decellularization combined with enzymatic removal of N‐linked glycans and residual DNA reduces inflammatory response and improves performance of porcine xenogeneic pulmonary heart valves in an ovine in vivo model

Limited availability of decellularized allogeneic heart valve substitutes restricts the clinical application thereof. Decellularized xenogeneic valves might constitute an attractive alternative; however, increased immunological hurdles have to be overcome. This study aims for the in vivo effect in sheep of decellularized porcine pulmonary heart valves (dpPHV) enzymatically treated for N‐glycan and DNA removal.


| INTRODUC TI ON
Despite great efforts to close the donation gap, shortage of human organs and tissues suitable for life-saving transplantations is likely to prevail in the future. 1 Therefore, research on alternatives is crucial in order to rescue many patients currently dying while waiting for a donor organ or suffering from suboptimal treatment (reviewed in 2 ).
One possible solution for this issue could be the use of animal-derived (xenogeneic) organs and tissues. Currently, there is a considerable progress in research efforts concerning xenotransplantation of solid organs. 3,4 In contrast to solid organs, xenogeneic tissues have already been used since decades for the generation of biological implants, like, for example, bioprosthetic heart valves. 5 However, since then, clinically applied xenogeneic substitutes basically had been chemically cross-linked, mainly by glutaraldehyde tanning, in order to prevent strong destructive host immune responses toward the graft tissues upon implantation. 6 Although showing very good hemodynamic properties and being the most often used heart valve replacement, chemically fixed xenogeneic valve bioprostheses feature specific disadvantages. Consequently, the quest for ideal heart valve prostheses is still open. 7 Currently under investigation in extensive multi-centric clinical trials, decellularized allogeneic heart valves potentially provide an optimal valve substitute, exhibiting long-lasting durability, physiologic hemodynamic properties, and the ability of adaptive growth processes. [8][9][10][11] Nevertheless, due to a substantial lack of human donor valves, an indefinite introduction of this innovative and promising regenerative therapeutic approach to the daily clinical routine turned out as being not possible in the foreseeable future. As a possible resolution approach, however, the simple transfer of decellularization strategies successfully applied on allogeneic heart valves to respective xenogeneic tissues approved to be unsuccessful yet, since decellularized porcine valve matrices hitherto always failed in clinical studies. 12,13 The poor performance of non-fixed xenogeneic tissues considered to be acellular in the past could be partially attributed to insufficient decellularization procedures. 14,15 Additionally, more recent in vitro experiments also revealed that the major-xenoantigen αGal as unique target of natural human antibodies constitutes a substantial part of porcine extracellular matrix proteins and thus cannot be removed by sole decellularization. 16 Essentially involved in the occurrence of severe immune mechanisms even against decellularized xenogeneic tissues once engrafted a αGal-negative recipient, αGal is no longer of substantial concern, since the generation of transgenic pigs lacking the αGal-epitope could be realized. 17 Decellularized heart valves derived from αGal-negative pigs, exhibit no differences in binding preformed human antibodies in comparison to decellularized human valves. 16 In contrast to preformed anti-αGal antibodies, which, without further xenogeneic stimulus, are present in high titer in all human, further xenoantibodies exist that exhibit variable, often low titer in healthy humans, but are rapidly elevated upon xenogeneic stimulus for, example in pig-to-baboon xenotransplantation experiments. 18,19 The two known antigens inducing such xenoantibodies are Neu5GC and the Sd(a) like antigen made by the porcine β-1,4-N-acetyl-galactosaminyltransferase 2 (β4GalNT2) (reviewed in 20 ). Using genetic engineering pigs lacking, these two antigens and also lacking αGal have been developed and showed reduced binding of preformed human antibodies. 21 Because all known xenoantigens are carbohydrates, we hypothesize that upon implantation mainly carbohydrates will be inducing new xenoantibodies. Therefore, the recently established enzymatic removal of N-linked glycans from decellularized matrices by applying PNGase F (will be published elsewhere) might be beneficial to achieve successful interspecies transplantation of decellularized heart valves matrices.
This issue was addressed in the current manuscript by first optimizing achievable decellularization extents for porcine pulmonary heart valves, second including enzyme-mediated removal of N-linked glycans and DNA, and third investigating the in vivo-performance of and the immunologic response toward respective decellularized porcine xenogeneic pulmonary heart valve conduits (dpPHV) orthotopically implanted into the established sheep model.

| Study design
Nine different detergent-based respectively enzymatic-supported decellularization protocols were employed to treat porcine pulmonary heart valves (3 animals per group) in order to identify the two most potent and suitable methods for the generation of structural intact, but cell-free grafts for heart valve replacement therapy. Per individual valve, each measurement (technical triplicates) was conducted at least 3 times per individual valve.
For further evaluation in vivo, the two selected protocols and the clinical applied Sodium Dodecyl Sulfate (SDS)/Sodium Deoxycholate (SD) decellularization were combined with DNase I and PNGase F treatment to remove residual nucleic acids and N-linked glycans. In total, 5 groups of dpPHV (n = 3, each, one animal died) were implanted for six months into sheep in orthotopic position: TX (TX: Triton X-100)

| Decellularization
In total, 49 (30 for in vitro and 19 for in vivo) fresh porcine pulmonary heart valves (pPHV) were excised from pig hearts obtained from a local abattoir. The genetic background of the animals is not defined, but based on oral information, the animals are German Landrace pigs. Because all animals have been of the same breed and originated from one farm, the genetic background of all animals used for this study should be comparable. Housing conditions have been identical, and all pigs slaughtered have been around 6 months old and about 90 kg of weight. pPHV were transferred on ice-cool PBS solution to the laboratory, extricated from excessive fat and connective tissue, and individually transferred to 250 mL laboratory flasks. Tissue disinfection was conducted in Braunol ® (B. Braun) for 5 minutes under continuous agitation, followed by a rinsing step of 20 minutes in PBS. Thereafter, pPHV were immediately subjected to decellularization ( Table 1). All incubation steps were performed on an orbital shaker at 200 rpm and at room temperature.

| Quantification of DNA, Hydroxyproline, GAGs, and SDS
Pulmonary artery (PA) samples were taken from the PHV to investigate DNA, hydroxyproline, GAGs, and SDS content. Matrix samples were minced with a tissue mortar after snap freezing in liquid nitrogen and resulting PA homogenates were freeze-dried. Per heart valve, at least triplicates of 2-10 mg dried matrix granulate were subjected to a complete Proteinase K (AppliChem) digestion. Conducted biochemical assays are described in detail elsewhere. 23 In brief, DNA was quantified using Hoechst 33258, 24

| Orthotopic implantation, echocardiographic examination and explantation
Animal experiments were performed in compliance with the Guide for the Care and Use of Laboratory Animals and approved by the local animal care committee of lower Saxony (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (LAVES), reference, 14/1527). The animal study was designed as an initial benchmark study with a small number of animals per group. Therefore, the group size (n = 3) does not allow a valid statistical analysis.
The surgical implantation was conducted as described by Theodoridis and Tudorache et al 9 and Lichtenberg et al. 27 In brief, after accessing the thoracic cavity via left lateral thoracotomy and establishing cardiopulmonarybypass according to standard procedures, PA was separated from the aorta and a truncal cuff approximately 2 cm in width was excised directly above the PA root in beating heart technique. Native pulmonary valve cusps were subsequently removed, and dpPHV orthotopically implanted into supravalvular position of female sheep (average weight 39.3 ± 1.2 kg) using running sutures for proximal and distal anastomoses. Transesophageal echocardiography was performed immediately after implantation as well as before explantation 6 months post-surgery. After sacrificing the animals, implants were excised from the explanted intact hearts ex vivo, macroscopically examined and subsequently separated into relevant component specimens, in order to be finally processed for histological as well as immunofluorescence analyses.

| Phalloidin and DAPI staining
For cell detection on the luminal side, half of one valve cusp of each explanted conduit was PFA-fixed in total and subsequently stained for cell bodies by using Phalloidin-Atto 488 (49409, Sigma-Aldrich) and nuclear stain by DAPI (Invitrogen) as previously described. 10,28

| Histological analysis and immunofluorescence staining
Representative parts of all explanted grafts were embedded in paraffin according to standard techniques. Standard hematoxylin and eosin (H&E) stain was conducted on 2-µm tissue sections of each sample as explicitly described before. 29

| Statistical analysis
Investigation of statistical significance was performed by two-way ANOVA followed by Bonferroni post-tests. Calculated numbers are given as mean ± standard deviation. A P value of ≤.05 is considered to be significant. P values are labeled as following: ≤.05 (*), ≤.01 (**) and ≤.005 (***).

| Influence of decellularization on DNA reduction, extracellular matrix proteins, and tissue retention of SDS
Biochemical analysis of decellularized porcine PHV regarding content of DNA, hydroxyproline, GAGs, and SDS revealed significant differences dependent on the applied decellularization protocol. In respect to residual DNA, the clinically approved SDS/SD protocol was unable to remove DNA, while the combined use of trypsin and TX was very efficient ( Figure 1A). Upon removal of cells and cellular debris, a direct impact of all types of decellularization procedures, the hydroxyproline content is increasingly measured in relation to the dry mass of the resulting cell-free matrix ( Figure 1B).
The analysis of GAGs shows that all decellularization processes remove GAGs, but dependent of the method applied the residual content may vary. Combining trypsin, TX, and SDS yielded the lowest values of remaining GAGs ( Figure 1C). The application of detergents for decellularization may result in detergent retention in the resulting matrix. Since detergents, especially SDS, are cytotoxic, we determined the SDS content depending on the decellularization method applied ( Figure 1D). Whereas, methods without SDS involved show no SDS content, and therefore, serve as negative controls, the TX and SDS method exhibited the highest concentration of matrix-related SDS, directly followed by the stand-alone use of SDS ( Figure 1D).

| Hemodynamic and clinical performance of decellularized porcine heart valves in sheep
Decellularized heart valves were successfully implanted into female sheep (n = 16) in orthotopic position (Figure 2A,B). One animal died during surgery due to uncontrollable arrhythmia. All others, either with allogeneic or xenogeneic implant, were in good physical conditions and showed a normal weight development from 39.3 ± 1.2 kg to 54.1 ± 5.1 kg during the 6-month experimental time ( Figure 2C).
However, in three animals, echocardiographic analysis revealed a massive dilatation of the pulmonary artery leading to graft failure ( Figure 2C). As these failures could not be assigned to a specific parameter, all three animals have been excluded from further analysis (highlighted in red, Figure 2C).  Figure 2C) a significant increase in mean valvular gradient and an increased degree of stenosis ( Figures 2C & 3B).
Interestingly, xenogeneic TX + SDS decellularized valves, additionally treated with PNGase F + DNase I, showed very good hemodynamic performance and no increase in the mean valvular gradient, or in insufficiency ( Figure 3B). The performance of all other xenogeneic heart valves was comparable to the allogeneic doPHV previously published ( Figures 2C & 3B). 22 Macroscopically, most xenogeneic explants exhibited thin, pliable, and translucent leaflets and a shiny and smooth PA comparable to allogeneic controls ( Figure 3A). Only TX + SDS-treated valves exhibited slightly reddish, thickened, and opaque cusps ( Figure 3A).
Furthermore, two out of three TX + SDS decellularized xenogeneic explants displayed signs of calcification, whereas no sign of calcification could be observed in any other explant.

| Histological analysis of explants
Conventional H&E stain revealed an overall intact histoarchitecture of all functional explants. Compared with all other groups, the cusps of xenogeneic TX + SDS-treated explants appear to be slightly thickened ( Figure 4A). All grafts have been covered by a highly vascularized adventitial layer. Cells could be found within the pulmonary artery as well as sinus region and the cusp. However, a complete repopulation with cells could not be seen, especially the distal part of the cusps happened to be cell-free in all explants ( Figure 4C). Applying an arbitrary score to judge, the cellular repopulation revealed the highest degree of repopulation for xenogeneic trypsin + TX+enzyme-treated valves ( Figure 4D). These valves and DNase I showed less inflammatory foci as their decellularized counterparts without enzyme treatment ( Figure 4B).

| Cellular repopulation of the cusp surface
Whole-mount staining with phalloidin and DAPI of half a cusp per valve revealed that the ventricular side was always better repopulated with recipient cells as the arterial side ( Figure 5A). Furthermore,  Figure 5B).

| Characterization of inflammatory cells found within the explants
Applying the pan-leukocyte marker, CD45 revealed that in all xenogeneic explants the proportion of CD45-positive cells was always bigger than in allogeneic explants ( Figure 6A-C). Among xenogeneic implants,

F I G U R E 4
Histological analysis of explanted heart valves. A, H&E staining of paraffin sections revealed an intact extracellular matrix of the pulmonary arteries and cusps. Considerable formation of neotissue caused the thickening of the cusps in xeno TX + SDS-treated valves. Cellular accumulations (inlet) composed of inflammatory cells (inflammatory foci) could be observed in all grafts at the side of the proximal anastomosis. Xeno TX + SDS and xeno SDS/SD implanted exhibited inflammatory foci at the anastomosis, but also within the sinus region and within the pulmonary artery, between tunica media and adventitia. B, Repopulating cells could be found in all valves within the pulmonary artery, cusp as well in the sinus region. Native-like repopulation of pulmonary artery and sinus region could be observed in trypsin + TX+enzyme-treated valves, whereas SDS + TX-treated allogeneic as well as xenogeneic valves exhibited incomplete repopulation of the pulmonary artery and sinus region. In all groups, heart valve cusps were only partially repopulated. C, Applying an arbitrary score to judge repopulation ( 9 maximum score = native =22), trypsin + TX+enzyme-treated valves exhibited the highest value, whereas allogeneic as well xenogeneic TX + SDS-treated valves showed the lowest degree of repopulation. D, Quantification of inflammatory foci revealed that xeno trypsin + TX+enzyme-treated valves exhibited similar numbers of foci as allogeneic implants. A high number of inflammatory foci was found in xeno TX + SDS implants, which, however, tends to be reduced in TX + SDS+enzyme-treated valves. vs TX + SDS+enzyme 20% ± 12% and SDS/SD 35% ± 10% vs SDS/ SD + enzyme 18% ± 8%). Phagocytic cells, like macrophages and some granulocytes, could be identified in all explants by staining with a CD11b antibody ( Figure 6D,E). No significant differences could be detected, although the proportion of CD11b-positive cells tend to be higher in xenogeneic compared with allogeneic explants ( Figure 6F). Interestingly, in allogeneic explants, about 10% of the invaded cells were CD11b, which is in the same range as CD45-positive cells ( Figure 6C,F). Consistently, almost no CD3-positive T cells (2% ± 1%) were found in allogeneic explants ( Figure 6G). On the contrary, a considerable number of CD3positive T cells could be found in xenogeneic explants ( Figure 6H). The number of T cells found in xenogeneic explants tends to be reduced by the enzyme treatment (TX + SDS 10 ± 5% vs TX + SDS+enzyme 6 ± 1% and SDS/SD 10 ± 6% vs SDS/SD + enzyme 7 ± 4%) ( Figure 6I).

| D ISCUSS I ON
In this study, we focused on the functional assessment and tol- As the DNA removal by trypsin + TX as well as TX + SDS decellularization per se resulted in low DNA content of the dpPHV, it is unlikely that the DNase I treatment lead to that positive effect.
PNGase F treatment, however, leads to a significant reduction in glycans present on decellularized porcine valves (will be published elsewhere). Thus, the PNGase F-mediated glycan removal from dpPHV might be responsible for the drop in CD3-positive T cells present in the explants ( Figure 6) and therefore indicative for increased tolerability.
In relation to previous studies investigating non-fixed porcine valves transplanted into sheep, the inflammatory reaction observed in our study seems to be much weaker. Ice-free cryopreserved porcine pulmonary valves tested by Biermann and colleagues exhibited intensive calcifications and a strong T cell-mediated immune reaction after 3 months in vivo. 32 On the contrary, other in vivo studies of decellularized porcine heart valves in sheep were reporting very good preclinical results. [33][34][35][36] Goldstein and colleagues reported successful implantation and function of SynerGraft valves made from decellularized porcine aortic valve tissue for up to 336 days in the sheep model. 33 Erdbruegger and colleagues also reported very good results of decellularized porcine heart valves implanted into sheep for up to 11 months. 35 Unfortunately, both, Synergraft and Matrix P prostheses, manifested a very poor outcome in clinical trials and failed due to massive immunologic reaction in man. 12,13,37 Therefore, the important questions are whether these encouraging results can be transferred to the human system and whether these results could justify further preclinical testing in non-human primates or even first clinical trials of decellularized xenogeneic heart valves?
Through thorough in vitro analysis, it became evident that these decellularized porcine valves failed due to incomplete decellularization. As antigens-like αGal epitopes are present on components of the insoluble extracellular matrix, a complete elimination of xenoantigens by decellularization is impossible. 13,15,16,38 Generating genetically based absence of αGal epitopes, however, results in abolishing of binding of preformed human antibodies to porcine decellularized heart valves. 16 This result and the availability of genetically modified pigs lacking not only αGal epitopes, but also neu5Gc and β2GalNT2 makes us confident that decellularized heart valves generated from such pigs will not undergo acute humoral rejection, when transplanted into humans.
A major limitation of this study is the low number of animals per experimental group, but in total, only three out of the 15 implanted xenogeneic decellularized heart valves failed. The failing was based on dilatations, most likely caused by suboptimal trimming and positioning of the graft during surgery resulting in unphysiological blood flow through the grafts. Remarkably, only three other valves showed the development of mild stenosis and insufficiency.
As shown in this study and already 24 before, decellularized matrices may retain SDS, when SDS was used as a decellularizing agent. Since SDS is cytotoxic, residual SDS, strongly bound to the matrix and potentially released during the remodeling process of the matrix in vivo, may interfere with the repopulation of decellularized matrices in vivo, and therefore, the lack of SDS in trypsin + TX decellularized heart valves compared with TX + SDS decellularized valves may account for a better outcome. As trypsin + TX-treated implants showed the best tolerability, it can be speculated that a higher repopulation could lead to less inflammation, mediated by an enhanced remodeling of the xenogeneic extracellular matrix.
The fact that an additional treatment of TX + SDS valves with PNGase F and DNase I considerably reduced the degree of inflammation, while not improving the repopulation, however, gives evidence that the two processes inflammation and repopulation can be separated. Both enzymatic treatments potentially help to remove xenoantigens. DNase I by destroying highly charged DNA molecules and PNGase F by removing N-linked glycans from glycosylated proteins being part of the matrix.
Since the known xenoantigens for humans, αGal and Neu5Gc are carbohydrates, we can speculate that others, which have not identified so far, may act as xenoantigens as well and the PNGase F treatment may has removed unknown xenoantigens resulting in less inflammation in the treated TX + SDS valves.
Conclusively, we could demonstrate that by efficient decellularization combined with enzymatic removal of DNA and N-linked glycans, the immunological barrier causing rapid destruction of xenogeneic tissues could be overcome.
Based on this study, trypsin + TX + DNase I + PNGase F treatment is the most promising approach that should be further investigated. Thus, extended in vivo observation periods, maybe up to several years, should be considered in the sheep model, while preclinical tests of decellularized xenogeneic heart valves in non-human primates such as baboons should be started.
The fact that trypsin damages the extracellular matrix must be kept in mind and the mechanical stability of the valves must be ensured prior to implantation. In this study, the trypsin treatment was limited to 90 minutes resulting in valves that showed excellent mechanical stability for 6 months in vivo.

ACK N OWLED G M ENTS
We want to thank Doreen Lenz for her excellent laboratory work and Astrid Diers-Ketterkat and Karin Peschel for crucial assistant during surgery. This work was funded by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) via the Cluster of Excellence "From regenerative biology to reconstructive therapy" (REBIRTH) and via project C7 of TRR127 (Biology of xenogeneic cell and organ transplantation-from bench to bedside) and the Cortiss foundation.

CO N FLI C T O F I NTE R E S T
Axel Haverich is shareholder of Corlife oHG, company producing decellularized human heart valves using the SDS/SD decellularization protocol described in the manuscript. All other authors declare to have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
RR, TG, KT, SS, KH, KF, AC, SC, and IT performed the experiments; RR and TG analyzed the DATA; AH and AH designed the experiments and secured funding; RR, TG, and Andres Hilfiker wrote the manuscript.