Objectives: Despite the advances in preclinical cardiac xenotransplantation, the immune reactions caused by species differences are not fully understood. Hyperacute rejection can now be avoided using genetically engineered donor organs, but cell-mediated rejection by the adaptive immune response has not been addressed successfully. Here we investigated the initial human pan-T-cell reaction using a pig-human blood working heart model.
Materials and Methods: Porcine wild-type hearts (n = 7) were perfused with human blood in a biventricular working heart system for 3 hours. As control, blood from the same human donors was circulated without a pig heart. Pan-T cells were selectively extracted from blood taken before and at the end of the perfusion cycle. The relative mRNA expression of selected target genes (real-time quantitative polymerase chain reaction) and the expression of microRNAs were determined.
Results: After xenogeneic organ perfusion, there was a moderate upregulation of several CD4+ marker cytokines (interleukin 2, interleukin 4, interferon ?) compared with control. We found a distinct increase in the mRNA expression of granzyme B and perforin, key markers of cytotoxic T cells. No differences in the marker genes of regulatory T cells were evident. Levels of the anti-inflammatory microRNAs miR-16 and miR-93 were significantly higher in the xenoperfused group than in the control group.
Conclusions: This study demonstrated that contact of human blood with pig endothelium activates cytotoxic T cells within the first few hours, indicating acute rejection processes. This is accompanied by upregulation of anti-inflammatory microRNAs, which may represent compensatory anti-inflammatory mechanisms.
Key words : Acute rejection, CD8+ cells, Granzyme B, MicroRNA, Perforin, Working heart model
Transplant is an effective therapy for patients with end-stage cardiac disease, but the number of organs available from deceased human donors is severely limited.1 Cardiac xenotransplant (CXTx) offers a possible alternative and solution to the shortage.2,3 Despite advances in preclinical CXTx, the immune reactions caused by species differences are not fully understood.2,3 The series of immune mechanisms that affect vascularized xenogeneic organs are commonly referred to as hyperacute rejection, acute humoral xenograft rejection, immune cell?mediated rejection, and chronic rejection.4 Hyperacute rejection is mediated primarily by the innate immune response and plays a much greater role in xenograft than in allograft rejection.5 Hyperacute rejection can be prevented using genetically engineered donor organs,4 but cell-mediated rejection mediated by the adaptive immune response has not been addressed successfully. In animal models, the role of T cells in xenograft rejection and the response to xeno-antigens is well recognized.6 For example, mouse helper T cells (TH cells) recognize disparate xeno-antigens as nominal antigens presented in association with self-molecules of the major histocompatibility complex (MHC), rather than directly as in alloantigen recognition.7 Graft survival after pig-to-baboon CXTx was significantly improved with genetically engineered pigs in combination with an immunosuppression regimen based on a nonhuman primate-mouse chimeric anti-CD40 monoclonal antibody.8 Costimulation blockade with anti-CD154 was more effective in prolonging survival than belatacept in a pig-to-primate renal transplant model.9 However, so far, human T-cell responses to cardiac xenografts have not been investigated in vivo.10
In this study, we investigated the initial human pan-T-cell reaction to CXTx in a pig-human blood working heart model.11 Porcine wild-type hearts were perfused with human blood for 3 hours, and then we determined the number of circulating CD4+ cells and CD8+ cells, analyzed the mRNA expression of marker T-cell cytokines, and quantified the inflammatory and anti-inflammatory microRNAs (miRNAs). Understanding the mechanisms that underlie immunomodulation will be crucial for successful clinical xenotransplant.
Materials and Methods
All animals received humane care in accordance with the Ludwig-Maximilians-University of Munich guidelines for the care and use of laboratory animals. The Government of Upper Bavaria approved the animal protocols. This study adhered to European and national law (Directive 2010/63/EU of the European Parliament and of the Council of September 22, 2010, on the protection of animals used for scientific purpose), and the protocols conformed to the ethical guidelines of the 1975 Helsinki Declaration.
Porcine wild-type hearts were cardiopleged and explanted. After 150 minutes of cold ischemia, the hearts (n = 7) were perfused with human blood ex vivo in a biventricular working heart perfusion system11 for a period of 180 minutes. For control experiments, the same perfusion system was set up without a pig heart, and blood from the same human blood donors circulated for 180 minutes. Blood samples were obtained before and after perfusion/circulation, and pan-T cells were isolated.
After intramuscular premedication with 10 mg/kg azaperone (Stresnil, Janssen-Cilag), 10 mg/kg ketamine (Ketavet, Pharmacia and Upjohn), and 0.02 mg/kg atropine (Atropinsulfat, B. Braun), healthy wild-type pigs were anesthetized with 1 mg/kg propofol (Disoprivan, 2%; AstraZeneca), and 0.01 mg/kg fentanyl (Fentanyl-Janssen, Janssen-Cilag) for midline sternotomy. Unfractionated heparin (Heparin-Natrium-25000-ratiopharm, Ratiopharm) was administered at 500 U/kg. The hearts were perfused through the aortic root by pressure-controlled infusion of 50 mL/kg cold (1-4 °C) cardioplegic solution (Custodiol, Dr. Franz Köhler Chemie). After explant, trachea and lungs were dissected from the heart, which was then placed on ice bath. This also allowed for cannulation of the heart in preparation for the perfusion system. Cold ischemia time was 150 minutes.
The pulmonary trunk, the aortic root (length, 50 mm; inner diameter, 11 mm), and both atria (length, 25 mm; inner diameter, 11 mm) were cannulated with custom-made cannulas. A 3F catheter (Arrow) was inserted through the vena cava caudalis into the coronary sinus. Electrodes were placed on the cannulas for continuous electrocardiogram monitoring. The heart was then de-aired, and the cannulas were connected to the perfusion apparatus.
For the pig working model, 500-mL samples of blood from healthy volunteer male donors were drawn directly before perfusion on the day of the experiment. Heparin at a concentration of 10 IU/mL was added to the donor blood. The blood was diluted with 6% hydroxyethyl starch (Volulyte, Fresenius Kabi) to achieve a constant porcine hemoglobin value of 8.5 mg/dL. The perfusion apparatus was first flushed with sodium chloride solution, and then 500 mL heparinized human blood was introduced into the system. Sodium bicarbonate and glucose were added to the blood perfusate to obtain a normal base excess between -2 and +2 mEq/L and a glucose concentration of 120 to 140 mg/dL. In the system, the blood was oxygenated until partial pressures of 100 to 150 mm Hg of O2 and of 35 to 40 mm Hg of CO2 were maintained. During the experiment, 400 mg/h of glucose and 0.2 IE/h of insulin were continuously infused into the system. For the control experiments, samples of the same amount of blood from the same donors were taken 3 months later, and samples were treated in the same manner as for the working heart model.
We used commercially available tubes and connectors that are known to provide biocompatible perfusion when used for cardiopulmonary bypass (ECC tubing, Raumedic). The ex vivo perfusion technique was as described previously.11 Briefly, human blood traveled first via a rotary pump (Deltastream, MEDOS Medizintechnik) from a cardiotomy reservoir and extracorporeal circuit tubing (ECC tubing, Raumedic) through a right heart bypass to an oxygenator (HILITE 800 LT, Rheoparin-coated hollow-fiber oxygenator, MEDOS Medizintechnik) and a 30-?m particle flow filter (Micro 40, Sorin Group) before reaching the heart. It was then infused retrograde into the aorta, the coronary arteries, and via the microcirculation of the heart into the coronary sinus. The right atrium finally led the blood flow back into the tubing system to the closed reservoir and thereby closed the circulation. During the reperfusion, the heart started beating. In case of ventricular fibrillation, the heart was defibrillated with 5 to 10 J from a manual defibrillator (Codemaster XL+, Hewlett-Packard). Retrograde flow through the aorta was maintained at a mean pressure of 65 mm Hg for 15 minutes. After this period, the bypasses were closed, and the Langendorff perfusion was switched to the working heart model.
Working heart model
First, the reservoir was slowly raised to a level so that preload of 6 mm Hg was reached in the right atrium. In consequence, the right atrium and ventricle were filled, and blood was ejected into the pulmonary artery during systole. By adjustment of a damping element, the pressure amplitudes in the artificial pulmonary circulation were reduced to physiological values of about 10 mm Hg. After passage through the oxygenator and microparticle filter, the blood filled the left atrium. A rotary pump helped to overcome the resistance of this artificial pulmonary circulation. The left ventricle then pumped the blood back
into the reservoir through the artificial systemic circulation. In this part of the model, 2 resistors and 1 damping element regulated the mean pressure at 65 mm Hg and the systolic pressure at 100 mm Hg during the entire course of the perfusion. Epicardial temperature was recorded to regulate the heaters of the reservoir and the oxygenator to ensure a mean heart temperature of 36 °C to 37 °C. An electrocardiogram was recorded continuously to visualize arrhythmias. Regular blood-gas analysis and measurement of activated clotting time helped monitor the acid-base balance and the coagulation system.
Blood samples were drawn from the pig working heart and from the control systems before (T0) and at the end (T1) of the perfusion cycle. Peripheral blood mononuclear cells were isolated by density gradient centrifugation (Histopaque 1077, Sigma-Aldrich) according to the manufacturer’s instructions. The total cell concentration and the percentage of viable cells were determined by the trypan blue exclusion method with a Vi-CELL cell viability analyzer (Beckman Coulter).
After cell count, T cells were selectively isolated using the MACS Pan T Cell Isolation Kit II and an autoMACS Pro separator (Miltenyi) according to the manufacturer’s instructions.
Flow cytometry and intracellular staining
All cytometric analyses were conducted using the BDTM LSR II flow cytometer (BD Biosciences). T cells were washed twice with phosphate-buffered saline, divided between 3 different tubes (0.2 × 106), and stained with LIVE/DEAD fixable blue dead cell stain (Invitrogen) to determine cell vitality and with fluorescent dye-labeled antibodies for surface expression of CD4, CD8, CD25, CD69, CD134, CD137, and CD154 (Table 1, Figure 1). Unstained and isotope controls were used to determine autofluorescence and nonspecific antibody binding. Multicolor compensation was performed with UltraComp eBeads (eBioscience).
RNA extraction, cDNA synthesis, and real-time quantitative polymerase chain reaction
RNA extraction was performed using the mirVana miRNA isolation kit according to manufacturer’s instructions. RNA concentrations were measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Equal amounts of RNA from the different time points were used to generate cDNA samples by reverse transcription using oligo-dT primers, random hexamers (Qiagen), deoxynucleoside triphosphates, RNaseOUT recombinant ribonuclease inhibitor, and SuperScript III reverse transcriptase (Thermo Fisher) according to the manufacturer’s instructions. All real-time quantitative polymerase chain reaction (RT-qPCR) analyses were performed in duplicate on a LightCycler 480 PCR system (Roche) with Universal ProbeLibrary probes and specifically designed primers (Metabion; Table 2) obtained through the Universal ProbeLibrary Assay Design Center (Roche). Cycling conditions were as follows: initial denaturation at 95 °C for 5 minutes, 45 cycles at 95 °C for 10 seconds, 60 °C for 30 seconds, and 72 °C for 15 seconds. Succinate dehydrogenase subunit A and TATA box binding protein were used as reference genes in all experiments. Relative mRNA expressions of target and reference genes were calculated using the Roche LightCycler quantification software.
Quantification of microRNA expression
Expression of miRNA was quantified with TaqMan miRNA assays (Applied Biosystems) according to the manufacturer’s protocol. The U47 RNA was used for normalization of miRNA expression data. Real-time PCR was performed in duplicate with the LightCycler 480 Probes Master on the LightCycler 480 instrument by applying the following cycling conditions: denaturation at 95 °C for 10 minutes, 45 cycles of 95 °C for 15 seconds, and 60 °C for 60 seconds.
After 3 hours of ex vivo perfusion, samples were taken from the left ventricle of the porcine hearts. Samples were fixed in 10% formalin, embedded in paraffin, and sectioned. Slides for histology and immunohistochemistry were stained with hematoxylin and eosin for light microscopy. Deparaffinized sections were prepared for immunohistochemistry. The primary antibodies were rabbit polyclonal sera against human immunoglobulin M (IgM; Dako, A0423) and human C4d (Emelca 1132588) diluted 1:50; the secondary antibodies were peroxidase-conjugated swine anti-rabbit antibody (IgM) or biotinylated swine anti-rabbit antibody (C4d). After incubation with avidin-biotin-peroxidase complex (C4d only) and diaminobenzidine hydrochloride, sections were counterstained with hematoxylin and mounted on coverslips.
All data were analyzed with GraphPad Prism 7.0 software and tested for normal distribution with the Shapiro-Wilk test. To determine the expression differences of the targets, values at time point T1 were related to the values at time point T0 and compared with the control group. For parametric distribution, differences between the groups were analyzed with the 2-tailed paired t test. In the case of nonparametric distribution, data were analyzed with the Wilcoxon signed-rank test. P values <.05 were considered statistically significant. In the case of parametric distribution, the data are presented as mean values (with SD) and otherwise as median values (with 25th and 75th percentiles).
Acute T-cell-mediated immune response
To determine the activation of T cells during CXTx, porcine wild-type hearts were explanted and subjected to 150 minutes of cold ischemia, then perfused ex vivo with human whole blood in a biventricular working heart perfusion model for 180 minutes. For control experiments, blood from the same human donors was circulated for 180 minutes in the same perfusion system without a pig heart. We compared the ratio of pan-T cells before and after perfusion. As shown in Figure 2a, there was no evident difference between the xenoperfusion and control groups (1.02 ± 0.1 vs 0.98 ± 0.04; P = .68). There was a high rate of cell death after the perfusion procedure, but this was similar between the xenoperfusion and control groups (1.71 ± 1.83 vs 1.23 ± 0.42; P = .81) (Figure 2b). Next, we used flow cytometry to analyze the ratio of CD4+ cells and CD8+ cells in each group, and this showed no significant difference in the ratio of the absolute numbers of the vital CD4+ cells (1.08 ± 0.16 vs 1.06 ± 0.14; P = .66) (Figure 2c) and CD8+ cells (1.02 ± 0.15 vs 0.97 ± 0.15; P = .37) (Figure 2d). In addition, T cells did not exhibit any differences in the expression of other surface antigens, such as CD25, CD69, CD134, CD137, and CD154 or in the T-cell costimulatory molecules inducible T-cell costimulator ICOS, T-cell immunoglobulin- and mucin-domain containing 3 TIM3, and the tumor necrosis factor receptor OX40. These results suggest that a xenoperfusion time of 3 hours may not lead to detectable alterations at the protein level. In further analyses, we thus focused on gene expression analysis.
mRNA expression of helper 1 and helper 2 T-cell cytokines
The T-cell support of immune responses falls into 2 broad categories: the generation of helper T cells (CD4+) and the generation of cytotoxic T cells (CD8+).12 CD4+ cells can be differentiated as different subsets:, helper 1 (TH1), TH2, TH9, TH17, and TH22 cells, regulatory T cells, and follicular helper T cells.13
First, we analyzed the mRNA expression of marker cytokines of CD4+ cells. A moderate upregulation in the xenoperfusion compared with the control group was detected for interleukin 2 (IL-2; 0.78 [25-75 percentile, 0.21-2.45] vs 1.29 [25-75 percentile, 0.61-5.77]; P = .047), interferon ? (0.74 [25-75 percentile, 0.40-3.05] vs 1.68 [25-75 percentile, 0.88-13.26]; P = .047), and IL-4 (0.78 [25-75 percentile, 0.12-1.13] vs 4.49 [25-75 percentile, 1.26-11.55]; P = .031), whereas no difference was measured for IL-10 (13.8 ± 12.88 vs 14.22 ± 10.47; P = .89). To test whether the xenogeneic perfusion may induce regulatory T cells, we analyzed the mRNA expression of the transcription factors forkhead box P3 and transforming growth factor ?. This revealed no difference between the xenoperfusion and the control groups.
We next investigated whether cytotoxic CD8+ cell responses may be influenced by xenoperfusion by quantifying the mRNA expression of granzyme and perforin, key mediators of the cytotoxic cell response. Xenogeneic organ perfusion resulted in significant increases in mRNA expression of both genes, ie, granzyme B (1.19 [25-75 percentile, 0.5-1.37] vs 2.27 [25-75 percentile, 1.90-11.88]; P = .015; Figure 3a) and perforin (0.96 [25-75 percentile, 0.52-1.33] vs 2.46 [25-75 percentile, 1.79-8.60]; P = .015; Figure 3b).
In addition to this cytotoxic CD8+ cell reaction, we found typical pathological findings for endothelial damage with an edema and hemorrhage. Examples for the histological and immunohistochemical findings of the myocardium after xenoperfusion are given in Figure 4.
Together, these results suggest that a short period of xenoperfusion activates cytotoxic T cells in human blood, whereas TH cells are marginally affected. In addition, we found endothelial damage in left ventricle examples.
MicroRNA expression of T cells
MicroRNAs are important regulators of almost all cellular processes. By targeting cellular signaling hubs, miRNAs have a fundamental regulatory effect on both innate and adaptive immune cells during inflammation.14 We quantified the expression of miRNAs known to be associated with inflammatory (miR-31, miR-150, miR-155, and miR-223) or anti-inflammatory responses (miR-16 and miR-93).14,15 No significant differences in miR-31, miR-150, miR-155, and miR-223 were detected between the xenoperfusion and control groups. In contrast, anti-inflammatory miR-16 (0.96 ± 0.24 vs 1.66 ± 0.74; P = .023) (Figure 5a) and miR-93 (0.89 [25-75 percentile, 0.43-3.9] vs 3.08 [25-75 percentile, 0.75-16.55]; P = .046) (Figure 5b) were significantly induced in the xenoperfusion group.
These results show that, after a short phase of CXTx, miRNA expression is altered toward an anti-inflammatory status.
Cardiac xenotransplant offers a potential alternative to allograft transplant in the case of end-stage cardiac diseases.2,3 However, clinical CXTx requires a robust means of overcoming the severe reactions mediated by both innate and adaptive immunity that limit the survival and function of xenogeneic organs.
Wilhite and colleagues have studied the T-cell response to wild-type and GGTA1-KO porcine cells in mixed lymphocyte reactions. They could demonstrate advantages of the elimination of the GAL epitope, the main target for preformed antibodies.16 The acute response of the pan-T cells with whole blood in physiological organ perfusion is uncharacterized until now. Here, we investigated the initial human pan-T-cell reaction after CXTx in a pig-human blood working heart model. After xenogenic organ perfusion of porcine wild-type hearts with human whole blood for 180 minutes, we found both a significant increase in the expression of molecules mediating cytotoxic T-cell responses and an induction of anti-inflammatory miRNAs.
In response to the inflammatory environment created by the innate response, cells of the adaptive immunity are stimulated to proliferate and to differentiate into cells with a range of functions appropriate for the immunological challenge.12 For the T-cell support of immune responses, there are 2 broad categories: generation of TH cells (CD4+) and generation of cytotoxic T cells (CD8+).12 We did not find differences in the ratio of pan-T cells or in the absolute number of CD4+ and CD8+ T cells before and after transfusion between the xenoperfusion and a control group. In contrast, Wilhite and colleagues showed after 4 days an increment of CD4+ as well as CD8+ T cells in wild-type and GGTA1-KO porcine cells in mixed lymphocyte reactions,16 which indicated that a longer observation period is needed to detect T-cell proliferation.
An overall high number of dead cells occurred after the perfusion procedure; however, there were no differences between groups, which indicated that cell death is caused by the mechanical stress and not by immune reactions. Xenoperfusion time of 3 hours did not lead to detectable alterations with respect to the expression of T-cell surface antigens. In addition, to acquire the effector functions necessary for donor rejection, naive CD4+ and CD8+ cells must be activated. This occurs by binding of the T-cell receptor to an MHC-peptide complex and by binding of costimulatory adhesion molecules on the T-cell surface with appropriate ligands on antigen-presenting cells.17 Our analysis of T-cell costimulatory molecules detected no differences in expression after xenoperfusion. Because the earliest markers of T-cell activation can be detected within 2 to 3 hours, peaking after 18 to 24 hours,18 longer perfusion times would be required to reveal such effects.
Because perfusion times in our model were limited by immune cell survival, we next analyzed mRNA levels, which respond rapidly to stimuli. Indeed, the contact of human T cells with porcine endothelium for 3 hours led to extensive changes in the mRNA expression of key molecules that mediate cytotoxic T-cell responses, even if the absolute number of CD8+ cells was not increased. Granzyme B and perforin are both markers of cytolytic proteins stored in granules of cytotoxic T cells that induce apoptosis in target cells upon release. Perforin facilitates the penetration of granzyme B into cytoplasm, whereas granzyme B induces cell death through caspase-dependent signaling pathways.19 Damage to the structural and functional endothelial integrity with edema and hemorrhage are present in our histological and histochemical samples. Immunoglobulin and complement deposition point to the well-described mechanism of antibody-mediated complement reaction.4 However, expression of granzyme B and perforin were significantly upregulated at the mRNA level during the experiment, which likewise explains the rapid induction of endothelial damage and acute cytotoxic rejection processes. These findings are in accord with acute rejection that occurs in allograft organ transplant.19,20 As already demonstrated for acute graft-versus-host disease, endothelial cells are a target of donor cytotoxic T lymphocytes, which causes endothelial damage consistent with our histology.21
For CD4+ cells, however, only a trend toward upregulation of marker cytokines was detected. Previous studies have investigated organ rejection after cardiac allograft transplant in animal models and humans and have shown an intense production of both IL-2 and interferon ? after several days and not within the first few hours of immune response.22-24
In recent years, there has been growing interest in miRNAs and their regulatory properties in the context of various diseases. MicroRNAs are promising biomarkers and attractive tools and targets for novel therapeutic approaches.25 In transplantation, miRNAs play an important role in graft rejection.26,27 Nevertheless, the miRNA expression profiles are poorly understood during the acute immune response after xenotransplant. We analyzed the expression levels of so-called “immune miRNAs,”14 and found significant induction of the anti-inflammatory miRNAs miR-16 and miR-93. The expression of these miRNAs was previously shown to be increased in stable coronary heart disease,28 which is in agreement with our findings. Both coronary heart disease and xenoperfusion of wild-type hearts cause endothelial inflammation, dysfunction, and ischemia due to the strong complement-mediated response to natural antibodies, which evokes a discrepancy between oxygen supply and myocardial oxygen demand.28 In both scenarios, upregulation of anti-inflammatory miRNAs could represent an intrinsic compensatory mechanism.
The design of an adequate control group was important to distinguish between effects caused by the perfusion apparatus and xenogeneic effects. Much attention was given to a delicate handling of the perfusate consisting of whole blood. At this point, only wild-type pig hearts were used in the perfusion group to characterize fundamental reactions in this model and the given time span. In the future the effects of genetic modifications can be tested using the described assays.
In summary, our study provides first evidence that the contact of human blood with pig endothelium may activate especially cytotoxic T-cell responses in the blood within the first few hours of xenoperfusion, which indicates acute rejection processes. Simultaneously, anti-inflammatory miRNAs are upregulated, most likely caused by compensatory reactions to limit inflammation. These altered miRNA expression patterns could possibly serve as diagnostic markers for early detection of graft rejection. However, detailed evaluation and validation are needed within the clinical setting of xenotransplant.
Volume : 19
Issue : 7
Pages : 708 - 716
DOI : 10.6002/ect.2020.0359
From the 1Department of Anesthesiology, University Hospital, Ludwig-Maximilians-University of Munich; and the 2Walter-Brendel Center of Experimental Medicine, Faculty of Medicine, Ludwig-Maximilians-University of Munich, Munich, Germany
Acknowledgements: This work was funded by the German Research Foundation (DFG), TRR127. Other than described, the authors have not received any funding or grants in support of the presented research or for the preparation of this work and have no declarations of potential conflicts of interest.
Corresponding author: Jan-Michael Abicht, Department of Anaesthesiology, University Hospital, LMU Munich, Walter-Brendel Center of Experimental Medicine, Faculty of Medicine, LMU Munich, Marchioninistr. 27, 81377 Munich, Germany
Phone: +49 89 4400 712170
Figure 1. Example of Cytometric Analyses of T Cells
Table 1. Fluorescent Dye-Labeled Antibodies Used for Cytometric Analyses
Table 2. Primer Sequences for Real-Time Quantitative Polymerase Chain Reaction
Figure 2. Acute T-Cell-Mediated Immune Response
Figure 3. Ratio of mRNA Expression of Marker Cytokines Granzyme B and Perforin of Cytotoxic T Cells Before and After Perfusion
Figure 4. Histologic and Immunohistochemical Examples of Porcine Wild-Type Hearts Show Typical Pathologic Findings After 3 Hours of Perfusion With Human Blood
Figure 5. Ratio of mRNA Expression of Anti-Inflammatories miR-16 and miR-93 Before and After Perfusion