Ecto-5′-nucleotidase (E5′N) is an endothelial surface enzyme that controls conversion of extracellular nucleotides into immunosuppressive adenosine. We evaluated whether expression of human E5′N on pig endothelial cells (EC) attenuates human NK cell-mediated cytotoxicity. A pig EC line was stably transfected with human E5′N and human NK cell adhesion and cytotoxicity toward pig EC cultures was measured by flow cytometry and intracellular enzyme release. E5′N activity in pig EC lysates increased from 0.68 ± 0.07 to 1013 ± 293 nmol/min/mg protein, whilst the rate of AMP to adenosine metabolism by intact cells increased from 0.37 ± 0.05 to >300 nmol/min/mg protein in non-transfected and transfected cells, respectively. The rate of adenosine production in transfected cells increased also with ATP as the extracellular substrate. Cytotoxicity of human NK cells was reduced from 10.7 ± 0.4% and 11.1 ± 1.1% with non-transfected pig EC to 5.2 ± 0.2% and 5.0 ± 0.2% in transfected cells with 50 μM and 250 μM AMP, respectively. Reduction of cytotoxicity in E5′N-transfected EC was abolished by the E5′N inhibitor and was mimicked in non-transfected EC by the addition of adenosine, demonstrating the key role of adenosine produced by E5′N in inhibiting NK cell cytotoxicity. We suggest that overexpression of E5′N in EC of transgenic pigs is a possible strategy to ameliorate rejection after xenotransplantation.
The use of xenogeneic cells, tissues and organs is one possible solution to circumvent the shortage of human organs for allotransplantation. Pigs are considered as optimal candidates for xenotransplantation not only because of similar anatomy and physiology, but also for economical and ethical reasons. Transplantation of pig organs into primates results in graft hyperacute rejection within minutes, thought to be due to the presence of xenoreactive natural antibodies (XNA) against the antigen present on the surface of pig cells, Gal-α(1,3)-Gal (1). This problem has been partially resolved in transgenic pigs expressing human complement regulatory proteins such as human decay accelerating factor (hDAF) or by knocking-out α1,3-galactosyltransferase (2,3). However, it is now clear that another process, known as delayed xenograft rejection or acute vascular rejection, occurs within days to weeks after pig-to-primate organ transplantation (4). This process is mediated by a broad spectrum of immune and hemostatic mechanisms involving antibodies, natural killer (NK) cells and monocytes.
NK cells have been implicated as mediators of delayed rejection in both rodent models of xenograft rejection (5–8) and in the pig-to-primate model (9,10). For example, in the pig-to-primate model, transplantation of kidneys (9) or hearts (10) results in a large NK cell infiltrate. Likewise, NK cells were the predominant cells infiltrating concordant hamster-to-rat skin (11) and heart transplants (12), discordant guinea pig-to-rat heart transplants (13) and a rat-to-human ex vivo heart model (14). Ex vivo perfusion studies of pig hearts with human blood suggest that the combination of NK cells and XNA are the major cause of organ injury (14).
Adenosine is of particular interest in the field of cardiac transplantation due to its combined cytoprotective, immunosuppressive and antiinflammatory effects. Adenosine exerts cardioprotection by vasodilation (15), antagonism of the chronotropic and inotropic effects of catecholamines (16), promotion of endothelial barrier function (17), inhibition of platelet aggregation (18), inhibition of TNFα (19), reduction of the complement component C2 level (20) and inhibition of neutrophil adhesion and free radical generation (21). NK cells have long been shown to respond to adenosine. These effects are mediated by surface adenosine receptors and their triggering results in inhibition of granule exocytosis (22) and attenuation of tumor recognition/adhesion (23). Adenosine can be produced by intra- and extra-cellular mechanisms. Extracellular ecto-5′-nucleotidase enzyme (E5′N, CD73) is expressed on endothelial cells (EC) and lymphocytes (24), with the exception of NK cells (25), and is an important regulator of the immune system. E5′N converts pro-inflammatory adenine nucleotides into antiinflammatory and immunosuppressive adenosine. E5′N acts in concert with another ectoenzyme, ectonucleoside triphosphate diphosphohydrolase (E-NTPDase, CD39), that is responsible for breakdown of ATP to ADP and then to AMP. E-NTPDase has a well established role in the prevention of xenotransplant rejection (26). Recent preliminary studies in our laboratory demonstrated that the pig activity of E5′N is one order of magnitude lower than in human EC (27). This may disrupt thrombotic and immune mechanisms regulated by nucleotides and adenosine in pig organ vasculature. Therefore an increased ability of porcine EC to convert nucleotides to adenosine may be highly beneficial in a xenotransplanted heart during both reperfusion and normoxic conditions.
The aim of this investigation was to evaluate whether human E5′N could be functionally expressed in pig EC and whether increased breakdown of extracellular nucleotides into adenosine will protect the pig EC from lysis caused by human NK cells.
Materials and Methods
Culture of pig endothelial cell line
The porcine vascular endothelial cell line (PIEC, derived spontaneously from a porcine iliac artery endothelial cell culture, a kind gift from Dr. K. Welsh, Oxford, UK) were cultured in RPMI (Gibco) supplemented with 2 mM L-glutamine, 150 U/mL penicillin/streptomycin and 10% fotal calf serum (FCS) referred to as complete medium, all from Sigma, UK, and passaged weekly. None of the nucleotides or inhibitors used in these experiments adversely affected the EC as assessed by cell proliferation assays. The cell line was characterized as EC by staining with monoclonal antibodies against CD31, ICAM, VCAM, E-selectin and swine leukocyte antigens (SLA) class I and class II using flow cytometry. There was no significant difference in the level of positive staining for these molecules between non-, E5′N- or mock-transfected cells used in this study and all three cell types demonstrated typical cobblestone morphology. PIEC were 100% positive for CD31, approximately 10% positive for VCAM-1 and 5% positive for E-selectin, demonstrating they were EC.
Generation of stable transfectants
The human 2.0kb E5′N gene (obtained from Dr. J. Spychala, Ann Arbor) was subcloned into pcDNA3.1(+) vector (Invitrogen) containing a CMV promoter. PIEC were transfected using Effectene (Qiagen) according to the manufacturers′ instructions. Mock-transfected cells were transfected with the plasmid only. Cell cultures were enriched for E5′N-transfected cells by positive selection using an antibody to CD73 (clone 1E9, a kind gift from Dr. Linda Thompson, Oklahoma Medical Research Institute) and a secondary purified goat antimouse IgG3 (2.5 μg/mL) antibody coated onto Dynabeads (Dynal, Inc., UK). Cells were passaged in the neomycin analogue geneticin (1 mg/mL) (G418 sulfate, Gibco) and selected weekly until cells were >95% positive for human E5′N.
Measurement of E5′N activity in intact cells
Confluent EC cultures were washed with Hanks balanced salt solution (HBSS, 25 mM Hepes, pH 7.35) and incubated for 30 min in 1 mL HBSS with 5 μM deoxycorfomycin (DCF, amino deaminase inhibitor) at 37°C prior to addition of 10 μM amino adenosine (AA, adenosine kinase inhibitor) and extracellular AMP to cultures for 1 h or 24 h. Fifty microliter samples were analyzed for nucleotide catabolites by reverse phase HPLC as previously described (28). Protein content in cell lysates was estimated using the Bradford method (29).
Biochemical assays for enzyme activities in lysed cells
Confluent EC cultures were lysed in 1 mL homogenization buffer (20 mM TRIS/HCl, pH 7.0, 150 mM KCl, 1 mM EDTA, 1 mM dithiothreitol) for 10 min at 37°C. The lysates were assayed for E5′N, AMP deaminase (AMPD), purine nucleoside phosphorylase (PNP), adenosine deaminase (ADA), ecto-ADPase (CD39), ecto-ATPase (CD39 L1) and adenosine kinase (AK) as we have described previously (30). The reaction was initiated by addition of 50 μL cell lysate to 50 μL of substrate solution at 37°C, and terminated by addition of 50 μL of 1.3 M perchloric acid. Samples were neutralized with 3 M K3PO4 and centrifuged at 13 000 rpm for 3 min. The supernatant was removed and analyzed by reverse phase HPLC (28). Protein content was determined using the Bradford method (29). The enzyme reactions were linear with respect to both the amount of extract and the incubation time.
Purification of NK cells
Resting NK cells were purified by negative selection using the RosetteSep Antibody cocktail following the manufacturer's instructions (StemCell Technologies, Canada). Briefly, fresh human peripheral blood was obtained from volunteer donors by venipuncture, and 0.5% (w/v) EDTA was added as an anticoagulant. The blood was incubated for 20 min at room temperature with a cocktail of antibody complexes, diluted 1:1 in PBS/2% FCS and layered over Lymphoprep (Nycomed, U.K.) for density centrifugation. NK cells were washed twice in PBS, resuspended in complete medium and purity assessed by flow cytometry.
NK cell adhesion assay
EC were plated at 1 × 105 per well overnight at 37°C. Fresh NK cells were isolated from human blood and resuspended at 5 × 106 cells/mL. One micromole CFSE was added for 5 min, and then an equal volume of FCS for 10 min in the dark at room temperature. Cells were washed twice at 1800 rpm for 6 min and resuspended in complete medium. 2 × 105 NK cells were added to EC for 1 h at 37°C. EC were washed twice in PBS and detached using Accutase (PAA laboratories, Linz, Austria), which is gentler than trypsin, for 10 min at 37°C. Cells were analyzed for 1 min using an EPICS-XL flow cytometer (Beckman Coulter, Palo Alto, CA) and gated to omit EC. Maximum adhesion was taken to be 2 × 105 NK cells in the same volume of medium as the samples.
NK cell cytotoxicity assay
Cytotoxicity was measured by the release of lactate dehydrogenase (LDH) using the CytoTox 96 Non-Radioactive Cytotoxicity Assay (Promega, UK) according to the manufacturers′ instructions. Briefly, 1 × 104 EC were incubated per well of a 96-well flat-bottomed plate overnight at 37°C. Human NK cells were freshly isolated and added to EC at 10:1 in 200 μL medium for 4 h at 37°C. Cells were incubated with nucleotides or the E5′N inhibitor αβ-methyleneadenosine 5′-diphosphate (AOPCP) for 30 min prior to addition of NK cells. EC lysed in 0.1% Triton/PBS were used to determine maximum lysis. Fifty microliter of substrate was added to 50 μL supernatant in triplicate and incubated for 30 min at room temperature. Fifty microliter stop solution was added to each well and the absorbance read at 490 nm. The percentage cytotoxicity was calculated as: % cytotoxicity = 100 ×[(Experimental − Effector Spontaneous − Target spontaneous)/(Target Maximum − Target Spontaneous)].
Flow cytometry and antibodies used
Cells (1 × 105) were resuspended in 100 μL PBS and stained with primary antibody for 30 min on ice. The monoclonal antibodies used to check NK cell purity were against CD16 and CD56 (Cambridge, UK), CD31 (Serotec) or purified from hybridomas (anti-MHC class I, W6/32). EC were characterized using antibodies against CD73 (clone 1E9, gift from Dr. L Thompson, Oklahoma Medical Research Institute), SLA Class I and II (VMBD, Washington, USA) and ICAM, VCAM and E-selectin (gifts from Prof. D Haskard, Imperial College, U.K.). Cells were washed in PBS and stained with FITC-conjugated goat antimouse IgG3 secondary antibody (Dako) for 30 min on ice. Cells were washed in PBS and fixed in 400 μL of 0.5% formaldehyde (BDH, Essex, U.K.). An EPICS-XL flow cytometer (Beckman Coulter, Palo Alto, CA) was used to acquire 5000 events.
Values are presented as mean ± standard deviation. Statistical analysis was performed using the unpaired Student t-test to compare two groups or by one-way analysis of variance (ANOVA) followed by Student–Newman–Keuls test to compare more than two groups. A p-value of <0.05 was considered as a significant difference.
Enzyme activities in cell lysates
The activity of E5′N in several different pig and human endothelial cell types, as well as pig cells transfected with human E5′N, is presented in Table 1. These results confirm significantly lower activity of E5′N in both non-transfected PIEC and PAEC than in any human EC, although activity in PAEC was slightly higher than in PIEC. Transfected PIEC showed remarkably high E5′N activity, an increase of over 1000 times compared to non-transfected cells. Comparison of the activities of the other enzymes of nucleotide metabolism is presented in Table 2, showing that transfection with E5′N did not affect the activities of any other cytosolic or membrane enzymes.
Table 1. Ecto-5′-nucleotidase (E5′N) activity in lysed pig and human endothelial cells (EC)
Ecto-5′-nucleotidase activity (nmol/mg protein per minute)
Data shown are mean ± SD, n = 3 to 5.
*p < 0.05 versus all human cells.
#p < 0.05 versus PAEC.
$p < 0.05 versus all other cell types.
Human umbilical vein EC (HUVEC)
35.7 ± 2.49
Human capillary EC (HCEC)
33.0 ± 3.06
Human EC line
38.7 ± 5.34
Human aortic EC (HAEC)
41.1 ± 4.2
Porcine aortic EC (PAEC)
2.79 ± 0.36*
0.68 ± 0.07*,#
0.58 ± 0.04*,#
1013 ± 293*,$
Table 2. Activity of various adenine nucleotide metabolizing enzymes in lysed non-transfected and E5′N-transfected pig endothelial cells. Data shown are mean ± SD, n = 3 to 5. There is no significant difference in activity between non-transfected and E5′N-transfected endothelial cells for any of the enzymes studied
Non-transfected pig EC nmol/mg protein per minute
E5′N-transfected pig EC nmol/mg protein per minute
21.6 ± 2.3
20.8 ± 2.3
56.3 ± 14.9
56.0 ± 24
2.5 ± 0.3
2.9 ± 0.3
4.8 ± 0.2
4.8 ± 0.8
3.2 ± 0.1
3.2 ± 0.2
3.2 ± 1.3
2.6 ± 0.3
Extracellular AMP catabolism by intact cells
Analysis of the breakdown of 50 μM exogenous AMP added to intact cultured non-transfected and transfected cells is presented in Figure 1. There was a massive difference in the rate of AMP degradation by non-transfected PIEC compared to those transfected with E5′N. The rate of adenosine production from 50 μM AMP was 0.37 ± 0.05 nmol/mg protein per minute in non-transfected PIEC, whilst the precise rate could not be quantified in transfected cells since almost all the AMP added was converted to adenosine in transfected cells after 5 min of incubation in this experiment. It was also not possible in a separate experiment (not shown) with a 1 min sampling time. This indicates that in transfected cells the rate was >300 nmol/mg protein per minute. Long-term incubation of EC with 1000 μM AMP showed that all the substrate was converted to adenosine in 5 h by the transfected cells, while the non-transfected cells metabolized only 15% of the total AMP within 24 h (Figure 2). The calculated initial rate of metabolism of 1000 μM AMP to adenosine in this long-term incubation was 177 ± 9.6 nmol/mg protein per minute in E5′N transfected cells, while in non-transfected cells the calculated rate was 0.67 ± 0.06 nmol/mg protein per minute.
ATP hydrolysis and formation of adenosine by intact cells
Formation of adenosine from 50 μM exogenous ATP by non- and E5′N-transfected PIEC is presented in Figure 3. By 30 min, the concentration of adenosine was significantly higher in transfected cells than in non-transfected cells (p < 0.01). The calculated rate of adenosine production by the non-transfected cells was 0.296 ± 0.031, while it increased to 0.642 ± 0.050 nmol/mg protein per minute in E5′N-transfected cells. At higher concentrations of ATP, the difference in the rate of adenosine production become even more significant and was 0.166, 0.194, 0.342 nmol/mg protein per minute in non-transfected cells compared to 1.61, 1.85 and 2.23 nmol/mg protein per minute in transfected cells at 250, 500 and 1000 μM ATP concentrations, respectively.
NK cell adhesion to pig cells
Adherence of human NK cells to non-transfected, mock-transfected and E5′N-transfected pig EC was not significantly different after a 1 h incubation in the absence or presence of AMP (Figure 4, all results compiled). However, there was large variation in the results obtained with NK cells from different donors. In some experiments, increased adhesion in the presence of 250 μM AMP to E5′N-transfected and non-transfected EC was observed (n = 2), whilst with other donors there was either less adhesion (n = 2) or no difference (n = 2) between E5′N-transfected and non-transfected cells. Addition of the inhibitor of E5′N (AOPCP) did not affect the patterns of adhesion (not shown).
Human NK cytotoxicity toward pig cells
There was approximately 50% inhibition of cytotoxic activity of NK cells toward E5′N-transfected PIEC as compared to non-transfected or mock-transfected cells (Figure 5). Exogenous AMP added at 10–250 μM concentrations did not affect the degree of this inhibition, suggesting that the level of endogenously released nucleotides may already be sufficient to exert maximum effect. There was no difference in the cytotoxicity of human NK cells toward non- and mock-transfected EC, either with or without AMP.
Effect of adenosine and E5′N inhibitor on human NK cytotoxicity toward pig cells
To determine the mechanism of protection from cytotoxicity, NK cells were exposed to adenosine for 30 min prior to adding them to EC, or EC were exposed to the E5′N inhibitor AOPCP alone or to 50 μM AMP with 100 μM AOPCP prior to the cytotoxicity assay (Figure 6). Preincubation of NK cells with adenosine reduced the cytotoxicity of NK cells toward non-transfected cells, but did not affect cytotoxicity toward transfected EC. This demonstrates that exogenous adenosine produced by E5′N-transfected cells can inhibit NK cytotoxicity. Preincubation with the E5′N inhibitor, with or without AMP, abolished the protective effect of E5′N overexpression on NK cell-mediated cytotoxicity, demonstrating that the protective effect of E5′N expression is related to its enzyme activity.
We have established that expression of human E5′N in pig EC attenuated human NK cell-mediated toxicity against pig EC. This study highlights for the first time the potential benefits of upregulation of E5′N in xenotransplantation. We have demonstrated that human E5′N can be expressed as a functional enzyme on the surface of pig EC and this may ameliorate the relative deficiency of E5′N in pigs compared to humans. Our experiments have also shown that E5′N activity controls not only the rate of breakdown of extracellular AMP to adenosine, but also defines the rate of adenosine production from the entire extracellular ATP breakdown cascade. Furthermore, we have established that the protective effect was related to the enzyme activity of E5′N as it was abolished by addition of specific inhibitor.
In these experiments, human E5′N and not the pig enzyme was used, because pig E5′N has not been cloned. Furthermore, the human protein is also less likely to be immunogenic, although the structure of the protein is relatively well conserved in vertebrate species (31). Expression of human E5′N on the cell surface of the pig EC was confirmed by flow cytometry, massive increase in E5′N activity in cell lysates and increased ability of cell monolayers to degrade extracellular nucleotides; the procedure did not diminish the viability or growth rate of the pig EC.
It is well established that purine nucleotides can be released at sites of tissue damage or inflammation (32) and this release may be involved in modulation of the immune response. ATP is the predominant nucleotide released and is processed by a cascade of ecto-enzymes including E-NTPDase, an enzyme with a well-established protective role in xenotransplantation (26,33). The product of these reactions, AMP, is then converted to adenosine by E5′N. The importance of this process is related to the diverse and antiinflammatory effects exerted by nucleotides and adenosine: ATP, and to a greater extent ADP, triggers pro-inflammatory and pro-aggregatory mechanisms (34), while adenosine induces the opposite effect by attenuating inflammation and exerting an antiaggregatory effect (35). The transfected cells used in our study converted all of the 50 μM AMP in the cell culture medium into adenosine within 1 min, despite significant diffusion distances in the wells of the culture plate, suggesting that the activity of E5′N is the major factor controlling maximum rate of adenosine production from extracellular AMP in vivo. However, as the major nucleotide released into the extracellular space is ATP (32), it could be argued that the supply of AMP by E-NTPDase could be the limiting factor and that the activity of E5′N activity could be in excess. Results of the experiments with ATP as the extracellular substrate indicated that adenosine production was increased three-fold in E5′N transfected cells, showing that the AMP dephosphorylation step was still affecting the rate of adenosine production. This indicates that the rate of production of AMP by E-NTPDase exceeds the capacity of E5′N in non-transfected cells and that increasing this activity in pig cells will lead to increased adenosine production. Although the diffusion distances are much smaller in the vascular bed than in culture plate, the contact of surface enzymes with the substrate is reduced by blood flow in vivo. Therefore only general conclusions can be made from our experiments and exact nucleotide degradation kinetics should be studied using different models such as flow-through systems for cultured cells.
We have shown that porcine EC can be protected against human NK cell-mediated damage by overexpressing human E5′N on porcine EC. The question arises as to the mechanisms of this protective effect. The evidence suggests that enzyme activity of the transfected molecule is essential and direct effect on NK cells is more likely than an effect in EC. Thus the protective effect of E5′N was inhibited by AOPCP, confirming the dependence on enzyme activity and production of adenosine. Supporting this hypothesis, cytotoxicity was reduced by addition of exogenous adenosine directly to NK cells prior to their addition to non-transfected EC.
Demonstration of the predominant role of E5′N enzyme activity suggests that supplying more substrate should enhance the effects; however, no further attenuation of cytotoxic effect was observed with addition of exogenous AMP. It is possible therefore that an endogenous supply of the AMP was formed from nucleotides and released from cells and this was sufficient to induce the maximum inhibition of NK-mediated cytotoxicity.
We did not observe a clear effect of E5′N transfection on NK cell adhesion to EC in our experiments in the absence of added AMP. However, we noted a varying response in NK cell adhesion when exogenous AMP was added, in some experiments there was increased adhesion, while in others decreased adhesion was observed in transfected compared to non-transfected cells. The trend was consistent with NK cells isolated from a single donor but different with cells obtained from different donors. The E5′N inhibitor AOPCP has no effect on the results of these experiments. This indicates a significant individual variation in this aspect of cellular response. Since we observed this variation only with AMP added, it is likely to involve an interaction between adenosine receptors and nucleotide receptors, triggered by ADP and ATP that, as recently described, can be formed from AMP on the cell surface (36).
There is previous evidence that low micromolar concentrations of adenosine are capable of inhibiting granule exocytosis in murine NK cells (23). However, the literature does not consistently report that adenosine inhibits NK cell cytotoxicity. Dombrowski et al reported that adenosine inhibited cytotoxic capability of a human NK cell line against K562 targets (37), albeit at high concentrations, but a later study suggested that only NK cell proliferation and not cytotoxicity was inhibited by adenosine (22). Adenosine regulates lymphocyte function through interaction with several types of adenosine receptors (38,39). Some authors have shown that adenosine interferes with the recognition of tumor cell targets by NK cells by acting through the A3 receptor on the effector cell (40) whilst others showed that an A3 adenosine receptor agonist potentiates NK cell activity (41). However, most of the earlier studies indicate that the stimulation of NK cell A2 receptors and subsequent increase in cAMP production is the most important in the depression of NK cell function, including alterations in the levels of granzyme B and perforin (42) and inhibition of target cell lysis (43–45).
The other potential benefits of E5′N overexpression will include inhibition of platelet aggregation as adenosine is a well-known inhibitor of this process (46), a fundamental element of xenograft rejection. Adenosine also inhibits apoptosis by increasing the expression of antiapoptotic Bcl-2 and decreasing expression of pro-apoptotic Bax proteins through the A2A receptors in in vivo canine models of reperfusion injury (47). In addition, stimulation of adenosine receptors also enhances the release of NO from cardiac myocytes (48) and EC. All these effects are likely to be beneficial following heart xenograft transplantation.
In conclusion, the significantly higher levels of adenosine produced from nucleotides by E5′N transfected pig EC compared to non-transfected EC protected the porcine EC from lysis by human NK cells. Xenotransplantation could provide the ultimate solution for increasing shortage of human organs for transplantation and recent advances, including the production of Gal-α(1,3)-Gal knockout pigs revived the hope for a solution to fundamental problems. These achievements combined with application of the findings reported in this study could help to develop xenotransplantation as a successful clinical procedure.
We thank Dr. Linda Thompson for her kind gift of anti-CD73 antibody, Professor Dorian Haskard for his gift of pig antibodies against E-selectin and VCAM, and Dr. Jozef Spychala for providing human E5′N cDNA. RTS is an Honorary Senior Lecturer at the Department of Biochemistry, Medical University of Gdansk. This study was supported by the Magdi Yacoub Institute, Grant M. PAF D. M. 427/7303/2002 and F. A. R. 2003 - University of Milano-Biococca and the British Heart Foundation (FS/2001011).