Genetically encoded Ca 2+ -sensor reveals details of porcine endothelial cell activation upon contact with human serum

The activation of the endothelial surface in xenografts is still a poorly understood process and the consequences are unpredictable. The role of Ca 2+ -messaging during the activation of endothelial cells is well recognized and routinely measured by synthetic Ca 2+ -sensitive fluorophors. However, these compounds require fresh loading immediately before each experiment and in particular when grown in state-of-the-art 3D cell culture systems, endothelial cells are difficult to access with such sensors. Therefore, we developed transgenic pigs expressing a Ca 2+ -sensitive protein and ex-amined its principal characteristics. Primary transgenic endothelial cells stimulated by ATP showed a definite and short influx of Ca 2+ into the cytosol, whereas exposure to human serum resulted in a more intense and sustained response. Surprisingly, not all endothelial cells reacted identically to a stimulus, rather activation took place in adjacent cells in a timely decelerated way and with distinct intensities. This effect was again more pronounced when cells were stimulated with human serum. Finally, we show clear evidence that antibody binding alone significantly activated endothelial cells, whereas antibody depletion dramatically reduced the stimulatory potential of serum. Transgenic porcine endothelial cells expressing a Ca 2+ -sensor represent an interesting tool to dissect factors inducing activation of porcine endothelial cells after exposure to human blood or serum.

(ECs) with human serum. 1,2 Moreover, the quantitative relationship of antibody deposition on the endothelial surface and activation of the complement system could be identified as major drivers of HAR. 3 Very soon, it became evident that the genetic modification of the donor, that is, the pig, would be the most promising approach to overcome antibody-mediated damage of the endothelium. Initially, additive transgenes expressing human complement regulatory proteins (hCRP) were introduced into the porcine genome, 4 aiming at suppressing the complement activation cascade that is triggered by antibody deposition. Although significant improvement was achieved by combinations of hCRP in pig-to-primate experiments, 5 HAR was finally overcome by eliminating the most relevant xenoepitope recognized by preformed human antibodies on porcine EC, α1,3-galactosyl-galactose (aGal), by deleting the responsible gene GGTA1 (GTKO). 6,7 More recently, it was shown that expression of a human thrombomodulin (hTBM) transgene overcomes incompatibilities in the coagulation system between the human (or primate) bloodstream and the porcine endothelium 8 and additionally prevents inflammatory-induced activation of endothelial cells. 9 The combination of GTKO.hCRP.hTBM with sophisticated immunosuppression and prevention of ischemia/reperfusion injury prolonged survival of a porcine heart heterotopically transplanted into a baboon for 2.5 years. 10 Furthermore, it facilitated the survival of baboon recipients after fully life-supporting orthotopic heart transplantation for 6 months. 11 In another study, life-sustaining transplantation of porcine kidneys into macaques was achieved by a combination of only GTKO.hCRP, 12 whereas the hTBM transgene proved essential in cardiac transplantation. 13 In contrast, attempts to transplant other vascularized organs such as liver or lung trail far behind. 14,15 In both cases, the poor outcome has been at least partially associated with insufficiently controlled endothelial activation. 16,17 Thus, our general understanding of endothelial function or failure in xenografts is far from satisfactory. There is evidence that the endothelium experiences significant acute or chronic challenges even in the best possible current circumstances: (a) Although downstream activation of complement is inhibited by hCRP, the ongoing deposition of antibodies and activation of initial complement components on the surface of the vessel walls might still activate EC. 18 (b) The direct activation of EC by different leukocyte types [19][20][21] suggests that additional cellular players deserve attention. (c) The presence of pro-inflammatory cytokines was sufficient to activate endothelial cells in the absence of human or primate cells or serum. 22 (d) The protective function of Corline-Heparin Conjugate on GTKO.hCRP.hTBM pig EC indicates space for further improvement of the xenograft endothelium. 23 Finally, the diversity of endothelial function and its dynamic regulation has stimulated the definition of endothelial subtypes, their characterization by high-end tools, as well as their comparative analysis on a high-density data basis. 24 Considering these aspects, it is highly relevant to promote research to better understand the status of xenograft endothelium.
Due to the difficulties in assessing vessel structures in vivo, improved tools for in vitro examination have been developed, such as the establishment of endothelial-coated vessel-like structures under shear stress, that are perfused either under constant 25 or pulsatile flow. 26 Further attention has been paid to the dynamic processes occurring during angiogenesis 27,28 or the communication of endothelial cells with each other 29 or with other cell types. 30 Interestingly, the analysis of these complex 3D-like structures was limited to a post-experimental single-point readout. This is somewhat surprising, as the quantification of intracellular calcium levels by Ca 2+ -sensitive probes is an established tool and many different probes are commercially available. Although the loading of tissue such as murine aortas with fluorescent Ca 2+ -sensitive probes was informative, 31 the examination of three-dimensional endothelial structures in vitro is apparently not compatible with the short-term interference of Ca 2+ -probe loading.
As Ca 2+ -sensitive proteins might be a valuable alternative, we describe here the generation of pigs expressing Case12, a genetically encoded fluorescent Ca 2+ -sensor based on green fluorescent protein, 32 and present evidence that Case12 sensitively detects activation of EC by biologically relevant substrates. It is of note that only complement-inactivated human serum, but not antibody-depleted serum, triggered calcium influx into the cytoplasm of porcine EC, confirming that antibody depletion alone is sufficient to reduce the activation of endothelial cells. Animal experiments included the generation of genetically modified animals by somatic cell nuclear transfer. Human blood was taken from healthy volunteers.

| Generation of case12-pigs
For ubiquitous expression of the synthetic Ca 2+ -sensor, we used the previously described CAG-MCS-pA/ neo plasmid 33 and introduced the coding region of Case12 gene (Evrogen) via the BamHI/ NotI sites in the multiple cloning site ( Figure 1A). The generation of genetically modified primary cells was performed as previously described. 34 In brief, the plasmid was verified by Sanger sequencing and prepared endotoxin-free; the CAG-Case12/neo element was excised from the backbone and nucleofected into primary fetal fibroblasts. Nucleofected cells were seeded and kept under G418 selection. 35 Somatic cell nuclear transfer (SCNT) was performed according to our routine procedures 36 and activated embryos were transferred to synchronized foster sows. Pregnancies were monitored and birth was introduced as described before. 33 Transgene expression was detected by Western blotting. and horseradish peroxidase-coupled polyclonal goat anti-rabbit antibodies (Jackson ImmunoResearch). Bound antibodies were visualized using ECL reagent (RPN2106; GE Healthcare).

| Primary cells
Fibroblast cultures were established from ear samples taken at an age of 3 weeks as described elsewhere. 35 At an age of 5 months, transgenic pigs were sacrificed and aorta samples were taken for cultivation of primary endothelial cells as described elsewhere. 8

| Human serum, heat inactivation, and immunoabsorption
Human blood was collected from healthy volunteers into polypropylene tubes containing glass beads (S-Monovette, Sarstedt, Germany) and allowed to clot for 30 min at room temperature. After centrifugation at 2000 × g for 10 min at 4°C, the supernatant was collected, stored in aliquots of 1 mL at − 80°C, and only thawed once for experimental purposes. In the present study, sera from 3 different donors with different blood groups were used both individually and pooled. Details are given in the respective figure legends.
Complement was inactivated by heating the sera at 56°C for 30 min.

| Anti-αGal ELISA
The ELISA for detection of human IgM and IgG antibodies specific for the Bdi epitopes was performed as described previously. 38 In brief, ELISA plates (96-well NUNC MaxiSorp, NUNC AB) were

| CH50 assay
The activity of the classical pathway of the complement was analyzed using standard hemolytic complement (CH50) assay with sheep erythrocytes (Biomerieux). Sheep red blood cells were washed in veronal buffered saline (VBS++) at 1:30 dilution until the supernatant was clear. The erythrocytes were then diluted to achieve the density of 10 9 cells/ml and incubated with rabbit anti-sheep erythrocyte antibody (S1389; Sigma-Aldrich) for 20 min at 37°C. After washing away the unbound antibodies, the sensitized erythrocytes were resuspended in Alsever's solution (A3551; Sigma-Aldrich) and stored at 4°C overnight. The following day, the cells were washed again, and the density was adjusted to 10 8 /ml. Sera were then added to a transparent 96-well microplate (Nunc) and incubated with the sensitized erythrocytes at 37°C for 60 min. The reaction was stopped with PBS. As control, erythrocytes were incubated in veronal buffer (DGVB++) and the reaction was stopped with water (T100, 100% lysis) or PBS (T0, 0% lysis, background). Optical density of free hemoglobin was measured at 412 nm (Infinite M1000 spectrophotometer, Tecan). Lysis percentage was calculated as follows: (OD sample−OD T0) (ODT100−OD T0) × 100.

| Activation response of Case12 PAEC to different stimuli
The principal function of Case12 as a Ca 2+ -sensor was proved by stimulating transgenic PAEC with biologically relevant components. First, we challenged cells with different concentrations of ATP ( Figure 2A) and revealed a dose-dependent response reaching a saturation plateau between 100 and 300 µM. The procedure was repeatable several times, albeit with a shrinking signal intensity ( Figure 2B). Interestingly, the PAEC did not react to histamine ( Figure 2A). This is in contrast to primary fibroblasts, in which both ATP and histamine induced profound Ca 2+ -influx into the cells ( Figure 2C), although the increase of fluorescence signal was much slower in the latter.
More importantly, however, not all cells responded in the same manner to ATP stimulation, but revealed signal courses that were rather sequential and graduated ( Figure 3A, Video S1, Figure S1).
Apparently, within a given window of examination, cells typically responded in the following manner: Initially, a driver cell reacted and generally showed the brightest and most sustained signal, and then, the other cells reacted with some delay but did not reach the same fluorescence intensity as the driver cell. Often cells showed Ca 2+ -influx one after the other, as in a serial connection. Of note, the differential fluorescent signaling was confirmed when the signal intensities of single cells were measured quantitatively ( Figure 3B).
While most of the cells reacted within a very short time-frame to ATP, the peak altitude they reached was apparently different and the signal decline was variable, revealing different slopes and occasionally also a fluctuating decline of fluorescence.

| Effect of manipulated human serum on PAEC activation
Transgenic porcine EC reacted profoundly but transiently to ATP, with the signal in single cells returning to baseline level mostly within a few seconds ( Figure 3B). This was in contrast to the stimulation of Case12-transgenic EC with human serum, the standard challenge applied to EC in the xenotransplantation context (Video S2;  Figure 5D), albeit with some minor changes: First, heat inactivation showed a reduction in fluorescence intensity, but not in the number of cells responding, and second, the overall stimulatory potential of antibody-depleted serum was reduced, but to a lesser extent than in the 1/10 serum dilutions.

| D ISCUSS I ON
Our study illustrates that genetic modification of pigs is not only relevant for tailoring donor animals for xenotransplantation, but might also have value for producing reporter pigs to gain more detailed insight into physiological and pathophysiological processes.
Although one would nowadays consider generating such models by using state-of-the-art technologies such as gene editing, 39  A prominent finding was that not all PAEC behaved equally upon stimulation with ATP and they failed to react upon histamine application. A simple explanation for this might be the sequential loss of specific receptors during in vitro culture, but a similar diversity in the activation potential has also been shown for endothelial cells in ex-vivo vessel systems. 31 The apparent diversity of EC activation is consistent with the current interpretation of "endothelial cells" as an umbrella term for cell (sub)types coating the luminal side of a vessel. It has become evident that EC diversity not only refers to the different vessel types in distinct tissues, but also to differences between endothelial cells in a local context. A substantial clue was the initial discrimination of EC by the expression of classical markers such as TBM. 42 Since then, endothelial subtypes have been discriminated at a broader molecular level by single-cell transcriptome profiling. 24 Although the consequences of this unexpected diversity and the underlying regulatory circuits remain elusive, this is in line with the increasing acknowledgment of diversity of other cell types and their Ca 2+ -mediated signal transduction, such as epithelial cells. 43 These divergent properties of individual EC also cast doubt on the popular use of immortalized endothelial cell lines, including those derived from different endothelial sources in the pig [44][45][46][47] Although such cells might carry essential markers of EC and match hallmarks of endothelial function, they lack the diversity of primary cells. Furthermore, the immortalization process itself and extended in vitro cultivation will impact endothelial physiology. On the other hand, the use of immortalized cells in high-throughput assays is attractive, and in such assays it might be useful to equip them with a genetically encoded Ca2+-sensor. From a long-term perspective, however, we propose that isolating large batches of different primary cell populations from genetically modified pigs will better resemble the in vitro counterpart of the in vivo situation.
Another relevant aspect of our study was the finding that human serum had a much more pronounced effect on individual EC than ATP, which might be related to the complex composition of biologically active agents in the serum, compared to the unambiguous structure of ATP. The sustained activation of PAEC over minutes, however, is remarkable, when the fast, but transient Ca 2+ -influx after ATP stimulation is taken into account. Even more F I G U R E 2 Stimulating Case12 cells with biologically active components. (A) In PAEC, ATP showed a concentration dependent stimulation up to a saturation plateau reached at 100-300µM ATP, whereas stimulation with histamine did not result in Ca2+-influx. (B) Repeated stimulation of PAEC with 100mM ATP (stim1, stim2, stim3) resulted in a sequential decrease of the calcium influx into the cells. (C) In contrast, fibroblasts were stimulated by both, ATP and histamine, albeit for the latter the reaction appeared slower, indicated by a significantly lower slope of the signal. The slope was calculated at the inflexion point of the time course of the cellular calcium rise F I G U R E 3 Time course of Case12-PAEC stimulation by ATP. (A) Upon administration of ATP to the culture medium, pictures were taken at regular intervals. (B) When specifying signals from single cells, it appeared that the increase of the signal took place within a relatively short interval, whereas the peak heights reached were different and the decline of the signal followed different courses F I G U R E 4 Stimulation of Case12-PAEC with human serum. Both the overall time lapse and the single-cell signals revealed a diverse pattern of EC response on human serum, spanning highly responding and long sustained signals to almost no reaction surprising was the finding that antibody deposition on EC alone, but not the abundance of complement in the medium alone, is sufficient to activate endothelial cells. The reduced activation of EC in the presence of inactivated complement at higher serum concentrations supports the idea that the alternative and lectin-based pathways might play a relevant role in endothelial activation, 18 but we have to consider that a content of 20% serum in the culture medium is unusually high for in vitro experiments and might, thus, contribute some artifacts. We interpret our findings such that the classical, antibody-mediated activation of the complement system is the dominating factor in EC activation when challenged with human serum.
Although our data give insight into the process of endothelial cell responses to xenorelevant stimulation, it would be interesting to examine the impact of existing and proven approaches to prevent antibody-mediated rejection, such as the GTKO.hCRP.hTBM genetic modifications. In this context, however, it would not only be necessary to combine the genetically encoded Ca 2+ -sensor with the F I G U R E 5 Modification of human serum. (A) Cytotoxicity of serum was completely abolished by heat inactivation and (B) antibody depletion reduced the levels of anti-αGal IgM and IgG to a level < 5%. (C) Stimulation with 10% serum revealed that only antibody-depleted serum reduced the overall fluorescence intensity as well as the proportion of reacting cells. (D) When serum concentration was increased to 20%, the number of cells reacting on antibody-depleted serum increased, whereas the fluorescence intensity decreased, compared to untreated human serum triple modification, but it would be important to analyze such cells in 3D-culture assays under continuous or pulsatile flow, ideally not only with plasma, but also with whole blood. Eventually, such an innovative, complex, and high-throughput enabling setup may greatly facilitate the further improvement of vascularized xenogeneic organs or tissues.
In summary, we present here a powerful tool for evaluating effects that activate the porcine endothelium. In a proof-of-concept, we report that antibody depletion alone, but not complement inactivation, is sufficient to significantly reduce the activation status of endothelial cells in the presence of human serum.

ACK N OWLED G EM ENTS
This work was supported by funding from the German Research Foundation (CRC127: xenotransplantation from bench to bedside).
Work performed by RS and RR was funded by the Swiss National Science Foundation (310030_182264). We thank Prof. Nicolai Bovin from the Shemyakin and Ovchinnikov Institute for Bioorganic Chemistry, RAS, Moscow, Russia, for providing anti-Gal immunoabsorption material and coating conjugates for ELISA.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

AUTH O R CO NTR I B UTI O N
AW designed the study, performed genetic modification of primary cells, cultivated endothelial cells, and performed activation assays.
PK and EMJ performed activation assays. RS and RR performed antibody depletion, complement standard activity test, and anti-aGal ELISA. AB performed work on pigs and isolated tissue for primary cell cultivation. MK and BK performed SCNT and ET. EK provided Western blot data. BR and CK supported design of the study. EW supported design of the study and contributed to writing of the manuscript. NK designed the study, designed and constructed the modification vector and wrote the manuscript.