Telephone: 31-10-7034776; FAX: 31-10-7032793
Liver grafts contain a unique subset of natural killer cells that are transferred into the recipient after liver transplantation
Article first published online: 6 APR 2010
Copyright © 2010 American Association for the Study of Liver Diseases
Volume 16, Issue 7, pages 895–908, July 2010
How to Cite
Moroso, V., Metselaar, H. J., Mancham, S., Tilanus, H. W., Eissens, D., van der Meer, A., van der Laan, L. J. W., Kuipers, E. J., Joosten, I. and Kwekkeboom, J. (2010), Liver grafts contain a unique subset of natural killer cells that are transferred into the recipient after liver transplantation. Liver Transpl, 16: 895–908. doi: 10.1002/lt.22080
- Issue published online: 25 JUN 2010
- Article first published online: 6 APR 2010
- Manuscript Accepted: 28 MAR 2010
- Manuscript Received: 23 DEC 2009
In contrast to other solid organ transplantations, liver grafts have tolerogenic properties. Animal models indicate that donor leukocytes transferred into the recipient after liver transplantation (LTX) play a relevant role in this tolerogenic phenomenon. However, the specific donor cell types involved in modulation of the recipient alloresponse are not yet defined. We hypothesized that this unique property of liver grafts may be related to their high content of organ-specific natural killer (NK) and CD56+ T cells. Here, we show that a high proportion of hepatic NK cells that detach from human liver grafts during pretransplant perfusion belong to the CD56bright subset, and are in an activated state (CD69+). Liver NK cells contained perforin and granzymes, exerted stronger cytotoxicity against K562 target cells when compared with blood NK cells, and secreted interferon-γ, but no interleukin-10 or T helper 2 cytokines, upon stimulation with monokines. Interestingly, whereas the CD56bright subset is classically considered as noncytolytic, liver CD56bright NK cells showed a high content of cytolytic molecules and degranulated in response to K562 cells. After LTX, but not after renal transplantation, significant numbers of donor CD56dim NK and CD56+ T cells were detected in the recipient circulation for approximately 2 weeks. In conclusion, during clinical LTX, activated and highly cytotoxic NK cells of donor origin are transferred into the recipient, and a subset of them mixes with the recirculating recipient NK cell pool. The unique properties of the transferred hepatic NK cells may enable them to play a role in regulating the immunological response of the recipient against the graft and therefore contribute to liver tolerogenicity. Liver Transpl 16:895–908, 2010. © 2010 AASLD.
It is generally recognized that after clinical liver transplantation (LTX), the incidence of chronic rejection is lower than after transplantation of other organ grafts. Furthermore, in about 20% of LTX recipients, immunosuppressive therapy can be withdrawn without occurrence of graft rejection.1 Various animal models are spontaneously tolerant to LTX, even though they reject other organs.2, 3 Furthermore, cotransplantation of a liver allograft can prevent rejection of other organ grafts from the same donor.4, 5
The mechanisms responsible for this relative tolerogenicity of the liver have only been partially elucidated. A number of observations in animal models indicate that the immune cells present in the liver graft may play a relevant role in the induction of tolerance. With LTX, so-called passenger leukocytes from the donor are transferred into the recipient and can establish a condition of chimerism of variable proportions and duration.6, 7 Independent studies from different groups have shown that in rat models, acceptance of liver grafts is strongly associated with the presence and the abundance of these passenger leukocytes.2, 3, 8-13 Depletion of passenger leukocytes from liver grafts abrogates tolerance, whereas their reconstitution restores the organ's tolerogeneic potential.2, 12 One of the possible mechanisms by which migrating donor leukocytes promote acceptance of liver grafts is by inducing apoptosis of alloreactive T cells in recipient lymphoid tissues.2, 14 Also in humans, a few studies have shown that after LTX, a variable number of donor lymphocytes are transferred into the recipient and can be detected, in situ or in the recipient circulation, for variable time periods.6, 7, 15, 16 An early study by Rao et al.,17 that shows how cell migration and chimerism were observed more dramatically after LTX compared to other organ transplantations, suggested that this phenomenon may explain why the liver is more tolerogenic than the kidney. This observation concurred with the initial suggestion from Starzl et al., in an article from 1993, which proposed that donor hepatic leukocytes were responsible for the enhanced tolerogenicity of liver grafts.18 In brief, although there are multiple observations indicating that the transfer of donor leukocytes may positively influence liver graft survival, the specific donor cell types involved in the induction of tolerance are not yet defined.
We observed that donor-derived myeloid dendritic cells migrate from liver grafts into LTX recipients. However, these liver-derived dendritic cells appeared to be stimulators of allogeneic T cell proliferation and proinflammatory cytokine production, thus suggesting that they represent a major player in the induction of acute rejection, instead of acting as promoters of tolerance.15, 19
A cell type that is abundantly present within the hepatic lymphocyte pool is the natural killer (NK) cell. Although NK cells in peripheral blood account for 10%-15% of lymphocytes, hepatic NK cells comprise 30%-40% of all lymphocytes present in a normal adult liver. These cells, originally called “pit cells”, were first described by Wisse et al. in 197620 and later were functionally defined as liver-associated NK cells.21 Doherty et al. have shown that this population, resident in the liver sinusoids, is highly cytotoxic.22 Until now, most of the work that has been performed on alloreactive NK cells in transplantation was dedicated to the study of recipient NK cells rather than the donor-derived populations. Although NK cells are classically regarded as effector cells contributing to rejection of allogeneic grafts, recent studies have shown their involvement in graft acceptance.23-25 NK cells seem to be involved in tolerance induction by either targeting donor-derived antigen-presenting cells or recipient alloreactive T cells.25 Both mechanisms are actually not mutually exclusive, but rather are complementary, and may be dependent on the biological model analyzed and the local environment in which NK cells are activated.25 After allogeneic bone marrow transplantation between selected donor-recipient pairs, donor-derived NK cells have been shown to reduce graft-versus-host disease by killing of recipient antigen-presenting cells, to promote graft acceptance by killing recipient alloreactive T cells, and to prolong leukemia-free survival of patients by eliminating residual leukemic cells.26, 27
In addition to their well-known innate functions, there is emerging evidence of the extensive cross-talk between NK cells and the adaptive arms of the immune system.25, 28 Although they are part of the innate immune system, NK cells are now becoming known also for their capacity to modulate adaptive immune responses. Therefore, if in the heterologous settings of transplantation, donor NK cells—which are able to distinguish allogeneic cells from self—migrate into the recipient, they may be involved in modulating the host immune response against the graft.
In this study, we determined the extent and duration of donor-derived NK cell migration into recipients after clinical LTX, and compared this to the NK cell chimerism after renal transplantation (RTX). We characterized donor NK cells that detach from human liver grafts in terms of both immunophenotypical and functional properties and compared them with blood NK cells. Finally, we determined the expression of a number of key integrins and chemokine receptors which indicate the migratory potential of NK cells in liver graft recipients.
PATIENTS AND METHODS
Peripheral Blood and Liver Graft Perfusate Collection
To determine the numbers of donor-derived NK cells and CD56+ T cells in the recipient circulation, we selected 7 human leukocyte antigen (HLA)-A2− LTX recipients, and 7 HLA-A2− RTX recipients who had undergone transplantation with a graft from an HLA-A2+ donor. HLA-A2 was used as a marker to distinguish between donor and recipient cells. All LTX procedures considered in this study were performed from a postmortem donor. Peripheral blood samples were collected from LTX and RTX recipients immediately before and 1 day after transplantation. To evaluate the duration when donor lymphocytes were detectable in the host circulation, we selected five additional LTX recipients who satisfied the same conditions described above; peripheral blood samples were collected before transplantation and during each one of the following time intervals after LTX: 1-5 days, 6-10 days, 11-15 days, 16-20 days, 21-30 days, 1-6 months, and 6 months to 1 year. Peripheral blood mononuclear cells (PMBCs) were prepared by Ficoll density gradient centrifugation. For this first part of the study, isolated PBMCs were stored in liquid nitrogen and thawed right before usage. Perfusates were collected from human liver grafts during the back table procedure. Upon arrival at the hospital, grafts were perfused through the portal vein with 1-2 liters of University of Wisconsin (UW) solution to remove residual blood from the vasculature. Immediately before transplantation, the donor livers were perfused once more with 200-500 mL of human albumin solution. These latter perfusates were collected from the vena cava and were used to study hepatic NK cells. Mononuclear cells (MNCs) from fresh liver graft perfusates were isolated by density gradient centrifugation. For immunophenotypic and functional analysis of blood NK cells, PBMCs were isolated from healthy individuals. The Ethical Committee of the Erasmus MC approved the study protocol, and written informed consent was obtained from each patient.
Detection of Donor-Derived NK Cells and CD56+ T Cells
Donor-derived NK cells and CD56+ T cells were detected in the recipient circulation using a fluorescein isothiocyanate (FITC)-conjugated anti–HLA-A2 monoclonal antibody (mAb) (BD Biosciences, San Jose, CA), in combination with CD3-phycoerythrin (PE) (BioLegend, San Diego, CA) and CD56-allophycocyanin (APC) (Beckman Coulter Immunotech, Marseille, France) mAbs. Cryopreserved samples of recipient PBMCs were used for this first part of the study. Dead cells were excluded from analysis by using 7-amino-actinomycin D (7AAD) (BD Biosciences Pharmingen, San Diego, CA). Binding of an appropriate isotype-matched FITC-conjugated control mAb was subtracted from binding of the anti–HLA-A2 mAb. Analysis was performed using a FACS Canto II flow cytometer (BD Biosciences) equipped with BD FACSDiva flow cytometry software, version 6.1.1 (BD Biosciences). At least 1 × 106 events were acquired from each sample.
Phenotypic Analysis of NK Cells
Mononuclear cells (MNCs) from liver graft perfusates and from blood of healthy individuals, LTX donors, or LTX recipients were thawed in fetal bovine serum (FBS) and stained with mAb to assess differences in the immunophenotypic properties of the CD3−CD56+ NK cells. The following mAbs were used: CD107a-PE, CD16-Pacific Blue, CD56-PE, CD62L-APC, CD69-APC, anti-Granzyme A-PE, anti-Perforin-PE, CD14-PE, anti-α4-PCy5, anti-β7-PE, anti-CXCR4-PE (BD Biosciences, San Jose, CA); CD158a,h-APC, CD158b1/b2,j-APC, CD94-PE, anti-NKG2A-PE, anti-NKG2D-PE, anti-NKp30-PE, anti-NKp44-PE, CD11a-FITC, CD19-PE (Beckman Coulter Immunotech, Marseille, France); anti-BDCA1-PE (Miltenyi Biotec, Bergisch Gladbach, Germany); chemokine (C-C motif) receptor 7 (CCR7)-PE (R&D systems, Abingdon, UK); CD16-FITC, CD3-PerCP-Cy5.5 (ExBio, Prague, Czech Republic); anti-Granzyme B-PE (Sanquin, Amsterdam, The Netherlands); anti-CCR9-PE, anti-CXCR3-APC, anti-CCR5-PE, and anti-CCR1-PE (R&D Systems, Minneapolis, MN). The following mAbs were used as isotype-matched controls: mouse immunoglobulin G1 (IgG1)-PE, IgG1-APC, IgG2a-FITC, IgG2b-FITC, IgG2b,k-PE, IgG1k-PCy5, IgG2a-PE, and rat IgG2a-PE. Optimal dilutions of all mAbs used were established in preliminary experiments. Flow cytometric analyses were performed using FACS Calibur or FACS Canto II (both from BD Biosciences, San Jose, CA).
Isolation of NK Cells from Peripheral Blood and from Perfusates
CD3−CD56+ NK cells were isolated from fresh liver perfusate MNCs and from fresh PBMCs by negative selection, using the NK Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's protocol. Briefly, non-NK cells were first labeled with a cocktail of biotin-conjugated antibodies against lineage-specific markers, and subsequently with anti-biotin MicroBeads. In addition to this, during NK cell isolation from perfusate MNCs, CD15 MicroBeads (Miltenyi Biotec) were added for complete granulocyte depletion. The entire cell suspension was then loaded onto a magnetic cell sorting (MACS) LD column that was placed in a magnetic field and retained only the antibody-conjugated cells. All the reagents used for the isolation were diluted in calculated proportions of MACS buffer (phosphate-buffered saline from BioWhittaker, Lonza, Belgium, supplemented with 2 mM ethylenediamine tetraacetic acid (EDTA) and 2.5% bovine serum albumin both from Sigma, St Louis, MO). Purity of isolated NK cells and proportion of the 2 subtypes were evaluated by flow cytometry using CD3-PE, CD56-APC, and CD16-FITC mAbs. Viability was determined by flow cytometry using 7AAD.
The average purity of the enriched NK cell population was 99.6% ± 0.5%, whereas the average viability was 97% ± 1%.
Cytotoxicity and Degranulation Assays with the K562 Cell Line
Flow cytometry was used for a combined assessment of degranulation and cytotoxicity. The first was quantified by measuring the surface expression of CD107a, a molecule detectable on cells after degranulation, and the latter was assessed by carboxyfluorescein succinimidyl ester (CFSE)-labeling of the targets and quantification of cell death. Freshly purified blood or perfusate NK cells were cultured overnight at 37°C in Roswell Park Memorial Institute medium supplemented with 10% FBS (Hyclone, Logan, UT), penicillin (100 U/mL) and streptomycin (100 μg/mL; both from Gibco BRL Life Technologies, Breda, The Netherlands). K562 target cells lacking major histocompatibility complex (MHC) class I expression were labeled with CFSE (10 nM; Molecular Probes, Eugene, OR) and cultured overnight at the same conditions. The following day, K562 cells were coincubated with NK cells in an effector:target ratio of 1:1. CD107a-PE mAb, or alternatively IgG1-PE mAb, was added at the beginning of the culture to the wells to determine the proportion of NK degranulation. After 4 hours of incubation at 37°C, the cells were washed twice with PBS and then stained with CD56 and CD16 mAb. The proportion of viable K562 cells was determined by using 7AAD; to quantify their absolute number, a fixed amount of beads (CaliBRITE unlabeled beads; BD Biosciences, San Jose, CA) was added to each sample and measured by flow cytometry. For each well, the absolute number of living K562 cells was calculated as a proportion to the number of beads; the percentage of specific cytotoxicity was then estimated as follows:
Four replicas of each sample for each condition were measured for every assay.
NK Cell Stimulation for Cytokine Production
Cryopreserved purified NK cells (5 × 104 cells/well) were stimulated with combinations of phorbol myristate acetate (PMA; 250 ng/mL) and ionomycin (500 ng/mL), interleukin-12 (IL-12) (10 ng/mL) and IL-15 (100 ng/mL), IL-12 (10 ng/mL) and IL-18 (100 ng/mL), or IL-2 (100 ng/mL) and IL-21 (50 ng/mL). Supernatants were collected after 48 hours of culture and assayed in duplicate using enzyme-linked immunosorbent assay (ELISA) for production of interferon-gamma (IFN-γ; Invitrogen BioSource, Nivelles, Belgium), IL-10, and IL-13 (eBioscience, San Diego, CA).
All data are presented as means ± standard error of the mean (SEM), unless differently specified. The Mann-Whitney U test was used to analyze whether differences between unrelated groups were significant. For paired comparisons, the Wilcoxon signed rank test was used, alternatively, when appropriate, the paired t test was applied. A 2-sided P value < 0.05 was considered significant.
Donor NK and CD56+ T Cells Migrate into the Recipient After LTX, but not After RTX
PBMCs from HLA-A2− patients who received LTX or RTX from an HLA-A2+ donor were used to analyze migration of donor-derived CD56+CD3− NK cells and CD3+CD56+ T cells into recipients. Flow cytometric analysis of PBMCs obtained before transplantation confirmed the absence of HLA-A2+ cells in the recipient, whereas donor splenocytes showed complete HLA-A2 positivity. One day after LTX, donor-derived MNCs in the recipient circulation could be identified as a separate cloud of cells expressing HLA-A2. Within this population, NK cells were defined as CD3−CD56+HLA-A2+ and CD56+ T cells were defined as CD3+CD56+HLA-A2+. At day 1 after LTX, an average of 7.0% (range 1.4%-17.3%) of the circulating NK cell population was of donor origin (Fig. 1A). Donor-derived NK cells were characterized by a high proportion of the CD56dimCD16+ subtype (Fig. 1A,B) and a low activation profile (the early activation marker CD69 was expressed by 23.1% ± 5.8% of CD56dim cells and by 22.0% ± 6.5% of CD56bright cells), similar to those of circulating recipient NK cells. Donor CD56+ T cells were also observed in the circulation (average 12.6%; range 6.1%-31.3% of all circulating CD56+ T cells). Notably, donor-derived NK cells and CD56+ T cells were only detected in LTX recipients, whereas lymphocyte chimerism was below the detection level of flow cytometry (each lymphocyte subset being <0.1%) in RTX recipients (Fig. 1C).
The average time-span in which these donor-derived cells were circulating in the recipient was determined in 5 patients who underwent LTX. Relevant clinical data of these patients are shown in Table 1. Although the proportions of donor NK cells varied among different LTX patients, in all cases, these cells were detectable in the circulation for an average period of 15 days (Fig. 2A). In the case of donor CD56+ T cells, a few differences were observed. First, the cells were detectable in the recipient in a wider range of proportions varying between 1.1% and 13.2% of the total CD56+ T cell population (Fig. 2B). Second, in 2 of 5 patients, circulating donor CD56+ T cells were detectable in high numbers up to 16-20 days after LTX.
|Patient||Diagnosis||Induction Therapy (Anti-IL-2R mAb)||Maintenance Immune Suppression||Acute Rejection||Ischemia Times (min)|
|4||Post-alcoholic liver chirrosis||No||STE/AZA/CsA/END||No||839||96|
|5||Insulinoma with liver metastasis||No||STE/TAC||No||390||72|
Liver NK Cells Show Differences in Their Subtype Composition and Their Phenotypical Features Compared with Blood NK Cells
To characterize the donor NK cells that detach from liver grafts, we collected perfusates during regular vascular reperfusion of grafts with human albumin solution before transplantation. Perfusate MNCs contained, on average, 44% ± 12% NK cells. Classical NK cell subsets were gated on the basis of the surface expression of CD56 and CD16 (Fig. 3), according to the work of Cooper and colleagues.29 Among the NK cells that detached from liver grafts, the CD56bright subset was enriched as compared to its counterpart in peripheral blood: whereas in blood only 9.9% ± 3.2% of NK cells (n = 8) belonged to the CD56bright population, this subset accounted for 45.8% ± 6.5% of the NK cells in perfusate (n = 17). Both in perfusate and blood, the majority of CD56bright cells were CD16−, whereas all CD56dim NK cells expressed CD16.
Immunophenotypical analysis of other surface markers was then performed in a comparative analysis between blood and liver NK cells (Table 2). The activation marker CD69 was expressed on the majority (93.8% ± 2.6%) of CD56bright liver NK cells, whereas the same subset in blood showed a significantly lower expression (23.0% ± 3.6%), indicating a physiological activation status of NK cells detaching from the liver. To test whether this activated profile was caused by the exposure of hepatic cells to UW preservation solution during graft storage, we incubated PBMCs or total blood from healthy individuals in UW preservation solution for 24 hours at 4°C, mimicking the storage conditions of liver grafts before utilization. In neither of the 2 cases could NK cell activation be observed. In contrast, CD69 expression on NK cells was slightly decreased after preservation in UW solution (data not shown). To exclude that the activation of NK cells is an artifact related to the premorbid state of the donor, we tested the expression of CD69 on paired samples of fresh perfusate and donor blood collected at the time of LTX. The results confirmed a physiological activation of the CD56bright subtype of hepatic NK cells and a low expression of CD69 on samples of donor blood (for liver NK cells: 96.4% ± 1.3% of CD56bright cells CD69+, 7.6% ± 0.9% of CD56dim cells expressing CD69; for blood NK cells: 3.9% ± 0.5% of CD56bright cells CD69+, 5.5% ± 3.2% of CD56dim cells expressing CD69).
|Total NK Cells or Subset||Blood NK Cells||Perfusate NK Cells||P Value|
|CD69||total||26.6 ± 4.8||68.5 ± 4.3||*0.0007|
|CD56dim||27.4 ± 4.8||34.0 ± 4.8||ns|
|CD56bright||23.0 ± 3.6||93.8 ± 2.6||*0.0007|
|NKp44||total||0.1 ± 0.0||2.2 ± 0.8||*0.0006|
|CD56bright||4.3 ± 1.1||9.2 ± 3.5||ns|
|NKG2D||total||60.3 ± 5.2||72.5 ± 5.5||ns|
|CD56dim||62.1 ± 5.9||65.4 ± 8.1||ns|
|CD56bright||90.6 ± 3.7||96.4 ± 1.7||ns|
|CD94||total||28.2 ± 7.3||49.2 ± 7.2||ns|
|CD56dim||25.4 ± 7.5||32.9 ± 5.9||ns|
|CD56bright||90.3 ± 2.0||76.9 ± 5.6||*0.0207|
|NKG2A||total||49.7 ± 7.5||52.6 ± 5.1||ns|
|CD56dim||49.3 ± 7.4||41.0 ± 5.5||ns|
|CD56bright||90.9 ± 2.4||72.3 ± 6.1||*0.0148|
|NKp30||total||45.3 ± 6.4||34.3 ± 7.3||ns|
|CD56dim||50.4 ± 6.9||45.9 ± 8.0||ns|
|CD56bright||67.0 ± 3.3||50.9 ± 6.2||*0.0379|
|CD158a (KIR2DL1 KIR2DS1)||total||20.3 ± 3.6||12.7 ± 2.7||†|
|CD56dim||22.4 ± 3.6||24.4 ± 2.7||†|
|CD56bright||3.0 ± 0.5||4.7 ± 1.9||†|
|CD158b (KIR2DL2 KIR2DL3 KIR2DS2)||total||30.9 ± 4.3||17.1 ± 3.8||†|
|CD56dim||33.8 ± 4.2||28.9 ± 4.3||†|
|CD56bright||4.4 ± 1.0||9.0 ± 4.1||†|
|CD11a (Mean Fluorescence Intensity)||total||68.6 ± 5.6||111.4 ± 6.9||*0.0003|
|CD56dim||73.4 ± 7.0||92.7 ± 4.7||ns|
|CD56bright||40.3 ± 3.1||138.9 ± 6.3||*0.0003|
The major inhibitory receptors on NK cells are the heterodimeric receptor NKG2A/CD94, binding HLA-E loaded with leader peptides from MHC class I molecules, and part of the killer Ig-like receptors (KIRs) that recognize different MHC class I alleles.30 As in blood, CD56bright NK cells from liver are characterized by a higher expression of NKG2A/CD94, and by a lower expression of KIRs compared with CD56dim NK cells (Table 2). KIR analysis was performed using 2 mAbs commercially available, CD158a and CD158b. These mAbs are able to distinguish sets of KIRs that are specific for a certain HLA-C group; however, they do not allow discrimination between activating and inhibiting receptors. CD158a mAb recognizes KIR2DL1 (inhibitory) and KIR2DS1 (activating) which are both receptors specific for HLA-C group 2 alleles. CD158b mAb detects KIR2DL2, KIR2DL3 (both inhibitory), and KIR2DS2 (activating), which are all receptors that can bind ligands of the HLA-C group 1. Data reported in Table 2 on KIR expression are not meant to compare blood and perfusate samples, because differences in KIR expression largely reflect the genetic variability among individuals. Therefore, we compared the presence of these receptors on paired samples of blood and perfusate material collected from the same donor (data not shown). The results indicated that there is no significant difference between NK cells derived from the 2 compartments.
Concerning the activating receptors, on average, 60% of blood NK cells and 72% of liver NK cells expressed the C-type lectin receptor NKG2D (Table 2). Likewise, hepatic and blood NK cells expressed comparable amounts of NKp30, belonging to the NK-specific family of natural cytotoxicity receptors (NCRs). Interestingly, although part of the CD56bright NK cells in perfusates expressed the activation-induced NCR NKp44, the same receptor was not present on CD56bright NK cells in blood. This result confirms that the hepatic CD56bright NK cells are activated.
Expression of CD11a, the α-chain of the integrin LFA-1, was higher on liver CD56bright NK cells compared with the same subset in blood. This adhesion molecule is highly expressed on all hepatic lymphocytes, mediating their adhesion to ICAM-1 on Kupffer cells and liver sinusoidal endothelial cells,31, 32 further confirming the hepatic origin of perfusate NK cells.
Intracellular staining was used to detect the content of perforin and granzymes A and B as an indication of the potential cytotoxic capacity of NK cells. Most CD56dim NK cells in liver and blood expressed these cytolytic molecules, whereas the proportion of hepatic CD56dim cells expressing perforin was lower compared to its counterpart in blood (Fig. 4). However, in contrast to CD56bright NK cells in blood, the majority of hepatic CD56bright cells expressed granzymes A and B, and about 50% of them expressed perforin. These results indicate that liver CD56bright NK cells have all prerequisites to exert cytotoxicity, whereas in blood and lymphoid tissues this subset has consistently been described as poorly cytotoxic.30, 33
Cytotoxicity of Liver NK Cells Is Enhanced Although IFN-γ Production Is Comparable to Blood NK Cells
To compare the cytotoxic capacity of liver and blood NK cells, freshly purified NK cells either from liver perfusates or blood were coincubated for 4 hours with CFSE-labeled MHC class I-devoid K562 cells in an effector:target ratio of 1 to 1. No activating cytokines were added in the culture system in order to obtain observations not biased by any in vitro activation. When surface expression of CD107a was quantified as a direct measurement of degranulation activity, hepatic NK cells showed a similar response compared with blood NK cells (Fig. 5A). Notably, when the 2 classical NK cell subtypes were gated, the CD56bright liver NK cells showed the highest degranulation activity (data not shown). To asses the cytotoxic capacity of NK cells, numbers of CFSE-labeled K562 where quantified by flow cytometry after incubation in the presence or absence of the effector NK cells (Fig. 5B). The average killing capacity of hepatic NK cells was 2-fold higher than their blood counterparts (Fig. 5C), indicating that perfusate NK cells are highly cytotoxic.
Because NK cell stimulation by K562 cells does not induce production of cytokines, purified NK cells were cultured with different combinations of stimuli and supernatants were assayed for the presence of cytokines. IFN-γ was induced in some of the conditions tested: the highest amount of IFN-γ was produced in response to a combination of IL-12 and IL-18, and no significant difference was found between liver and blood NK cells (Fig. 6). Similarly, the combination of IL-12 and IL-15 induced significant amounts of IFN-γ and, as for the previous condition, large variability among samples was observed. By contrast both the stimulation by PMA and ionomycin and by IL-2 and IL-21 did not induce any IFN-γ production. Furthermore, both liver and blood NK cells were unable to produce IL-10 and IL-13 in all the conditions here tested (data not shown).
Donor NK Cells in the Recipient Express L-Selectin and the Integrin α4/β7
To estimate the homing potential of donor NK cells that detach from liver grafts, we measured the expression levels of a number of relevant chemokine receptors, selectin and integrins both on NK cells that detach from the grafts during reperfusion and on donor-derived NK cells found in the circulation of the recipient.
First, perfusate NK cells were analyzed (Fig. 7A). The chemokine receptors CCR1, CCR5, and CXCR3 involved in homing into inflamed tissues showed no detectable expression on liver NK cells (N = 4), suggesting that they have no capacity to migrating into inflamed tissues. Expression of the chemokine receptor CXCR4, selected to identify potential homing to bone marrow, was generally low. Around 15% of hepatic NK cells expressed the gut-homing receptor CCR9, and 38% expressed the two chains of the integrin α4/β7, a ligand for mucosal addressin cell adhesion molecule-1 (MAdCAM-1) which is expressed on the high-endothelial venules of Peyer's patches and sinus-lining cells of the splenic marginal zone. For both markers the expression was higher on the CD56bright subset than on the CD56dim. L-Selectin (CD62L) and the chemokine receptor CCR7, both involved in selective migration into secondary lymphoid organs, were detected on small proportions of both CD56dim and CD56bright liver NK cells. In contrast, in blood 37.6% ± 5.2% (n = 7) of NK cells are CD62L-positive. Moreover, although CCR7 is not expressed on peripheral CD56dim cells, 66.9% ± 4.0% of CD56bright cells in blood are CCR7-positive (n = 7).
Expression of homing receptors on migrating HLA-A2+ donor NK cells in peripheral blood obtained from patients shortly after LTX was comparable to that of circulating HLA-A2- recipient NK cells (Fig. 7B). High expression of α4/β7 characterized both donor (47.2% ± 9.4%) and recipient NK cells (35.8% ± 10.8%), whereas CCR9 was detectable but on a lower proportion of circulating NK cells. CD62L-positivity was observed on significantly higher percentages of circulating donor than perfusate NK cells (10.1% ± 3.8% on perfusate NK cells; 32.4% ± 8.3% on donor NK cells, P = 0.031; 36.5% ± 6.0% on recipient NK cells, P = 0.016). In summary, these data indicate that a large proportion of donor-derived NK cells that migrate in the circulation of LTX recipients express the α4/β7 integrin, allowing re-allocation into the recipient spleen, and a minority coexpress CCR9, allowing homing to gut-associated lymphoid tissues.
This study demonstrates that a large proportion of human hepatic NK cells belong to the CD56bright subset, and exhibit a physiologically activated profile. Our in vitro functional assays indicate that they can kill their targets more efficiently than their counterpart in blood and secrete IFN-γ, but not Th2-cytokines or IL-10, upon stimulation with monokines. During LTX these hepatic NK cells are transferred into the recipient and their unique properties may enable them to play a role in regulating the immunological response of the recipient against the graft by exerting allo-reactivity against recipient immune cells entering the liver. Moreover, we show that after LTX, but not after RTX, significant numbers of donor NK cells (mainly belonging to the CD56dim subset) and CD56+ T cells mix with the recirculating pool of recipient NK cells for approximately 2 weeks. The donor NK cells in the recipient bloodstream express chemokine receptors and integrins that indicate that a minority has the potential to be relocated into spleen or gut-associated lymphoid tissues of the recipient.
It is not known what determines the release of donor lymphocytes from the liver graft during perfusion, but one major trigger has been suggested being the cold preservation of the liver before the graft is transplanted. The preservation conditions can determine an injury to the adhesion molecules of sinusoidal-lining cells thus favoring the release of viable cells into the sinusoidal lumen.34, 35 This hypothesis is supported by the demonstration of the large number of leukocytes—ranging between 106 and 108—in the liquid collected after reperfusion of the liver at the end of the cold storage period (our data and Bosma et al.,15 Demirkiran et al.,16 and Jonsson et al.36). Although this might represent the main mechanism triggering the release of hepatic leukocytes, no clear correlations were found between the grade of chimerism and the duration of cold or warm ischemia times in the cases considered in this study (data not shown).
Our data confirmed that liver NK cells are phenotypically and functionally different from the correspondent cell type in blood. First, hepatic NK cells are enriched for CD56bright cells, which represent about 50% of the total liver NK cell population. By contrast, in blood the CD56bright cells account for about 10% of the total population. Second, although it was already known that intrahepatic lymphocytes are characterized by a high activation state,15, 36-39 we have shown here that it is specifically the hepatic CD56bright subset that expresses the early activation marker CD69 and the activation-induced NCR NKp44. We excluded that this liver-specific activation could represent an artifact due to the premorbid state of the donor or due to the exposure to low temperatures (4°C) or UW preservation solution. Similarly, other authors have already shown that the isolation procedure of perfusate leukocytes, including prolonged exposure to cold UW solution, did not promote activation of mononuclear cells nor did it enhance their stimulatory capacity in MLRs.36, 38
Third, CD11a expression was elevated on the CD56bright subset. CD11a is an adhesion molecule highly expressed on all hepatic lymphocytes mediating their adhesion to ICAM-1 on Kupffer cells and liver sinusoidal endothelial cells.
Functionally, freshly isolated liver NK cells killed K562 target cells more efficiently than blood NK cells. This effect could be observed without any stimulation by external cytokines and at a very low effector:target ratio. Although several studies in rodents have already demonstrated that hepatic NK cells are more cytotoxic than the same cell type in blood or spleen,40-42 few studies describe the cytotoxic properties of human liver NK cells.22, 43 The group of O'Farrelly reported that NK cells isolated from biopsy specimens from normal human liver tissue were able to kill K562 target cells but that high effector:target ratios were necessary: clear cytotoxicity was distinguishable from a 50:1 effector:target ratio.22 This difference from our results may be attributed to different methods used for quantification of cytotoxicity, ie, the conventional chromium-release assay, which is known to be less sensitive compared to the flow cytometric technique used in the present study. Similarly, the group of Asahara showed that IL-2 activated human liver NK cells were more cytotoxic against an hepatocellular carcinoma cell line compared with blood NK cells; however purified and unstimulated liver NK cells were not tested in these experiments.43 Our results demonstrate that freshly isolated NK cells from human liver kill target cells more efficiently compared to their counterparts in blood, without the need of prior activation in vitro. Although in blood the CD56bright subset of NK cells is mainly considered to produce cytokines,29, 30, 44-46 we here demonstrated that in liver this same subtype shows clear cytotoxic degranulation, as measured by detection of CD107a on their surface, in response to target cells. Additionally, intracellular staining of the 3 most common cytolytic enzymes—perforin and granzymes A and B—showed a higher proportion of hepatic CD56bright cells containing these molecules compared to the same subtype in blood. In addition, liver NK cells secreted IFN-γ but not IL-10 nor IL-13 upon stimulation with monokines.
Hepatic NK cells differ also from NK cells resident in lymph nodes.33 Although an enrichment of the CD56bright subset has been described also for lymph nodes, these cells show a resting immunophenotype (CD69negative), do not express the NCRs NKp30 and NKp44, are perforin negative and are unable to kill target cells; furthermore IL-2 stimulation is required for induction of perforin expression and cytolytic activity.33
By LTX liver-resident leukocytes are intrinsically transferred from the donor into the recipient, leading to the establishment of a variable grade of chimerism that may positively affect the induction of graft tolerance.6, 7, 34, 47, 48 In this study we show that, in addition to T cells, B cells and MDC,7, 16, 19, 34 donor-derived NK and CD56+ T cells migrate into the recipient circulation. Although previous studies have already shown that CD56+ cells7 and CD16+ cells34 are transferred into recipients after LTX, the authors did not discriminate between NK and CD56+ T cells. Our data show that all LTX recipients are chimeric in NK and CD56+ T cell lineages for about 2 weeks after transplantation. In contrast to NK cells detaching from liver grafts during pretransplantation perfusion, the majority of donor-derived NK cells that mix with the recirculating pool of recipient NK cells belonged to the CD56dim subtype. This discrepancy can be explained by two distinct hypotheses: first, the CD56bright subset may be preferentially retained within the liver sinusoids by adhesion molecules, such as CD11a, which is highly expressed on this hepatic subset (Table 2). Alternatively hepatic CD56bright NK cells, which are highly activated and have a higher expression of the α4/β7 integrin, rapidly relocate from the recipient circulation to other compartments, such as gut-associated lymphoid tissue, thus determining their absence from the bloodstream as early as few hours after LTX.49
The fact that donor lymphocyte chimerism can be more or less prominent among different patients is influenced by a number of factors.34 The total number of cells released from the liver, the rate of clearance of the allogeneic cells and their migration to other tissues are all aspects that can play a significant role in this context. A previous study excluded that allogeneic lymphocytes migrating in recipient circulation after LTX are derived from blood transfusions administered during the transplantation procedure, thus confirming that leukocyte chimerism after LTX is due to the intrinsic transfer of such cells by liver transplantation procedure.7
Analysis of selectin, integrin and chemokine receptor expression on perfusate and migrating donor NK cells gave few indications of where these cells may be homing after detaching from the donor liver. No detectable expression was observed for 3 chemokine receptors, CCR1, CCR5 and CXCR3, involved in lymphocyte homing into inflamed sites.50-52 Although CXCR4, the receptor for homing to bone marrow,53, 54 was detectable at low levels on perfusate NK cells, donor-derived NK cells in the recipient circulation were CXCR4negative, suggesting that these cells are not homing to bone marrow. CCR7 and L-selectin (CD62L) were used to test the potential migration properties to lymph nodes. Although CCR7 is a receptor responding to CCL19 (MIP-3-beta) and CCL21, CD62L can bind to addressins on high endothelial venules to facilitate extravasation of lymphocytes from the circulation. Notably, whereas expression of both molecules was low on perfusate NK cells, we detected a significantly higher expression of CD62L on donor NK cells circulating in the recipient, but CCR7 remained low, suggesting that migrating donor NK cells have limited potential to enter recipient lymph nodes.
Under normal conditions expression of the ligands for CCR9 (CCL25) and for α4/β7 (MAdCAM-1) is restricted to the gut.55, 56 Therefore, the minority of migrating donor NK cells that coexpress CCR9 together with the integrin α4/β7 may selectively home into gut-associated lymphoid tissue.55, 56 Interestingly, homing of donor lymphocytes to gut-associated lymphoid tissue after LTX was recently observed in an animal model.49
Finally, whereas in animal models donor NK cells have been found to migrate into the host spleen after LTX,57 in humans the mechanism of leukocyte homing to spleen is not yet completely unravelled. Although localization of lymphocytes in the red pulp of the spleen has been described as a passive process,58 migration into the white pulp is independent of interactions with high endothelial venules but rather involves the expression of MAdCAM-1 by the sinus-lining cells.59 Therefore, the 40%-50% of migrating donor NK cells that expressed the α4/β7 integrin but did not coexpress the gut-homing receptor CCR9, may be able to enter the white pulp of the recipient spleen.
Although the functional role of donor lymphocytes transferred into the recipient is yet not clear, most of the studies indicate that they have a protective role against rejection of the transplanted liver. This hypothesis is clearly supported by numerous studies in animal models in which spontaneous tolerance to liver grafts, which is observed in a number of fully allogeneic rat strains combinations, can be abrogated by reduction of passenger leukocytes.2, 12, 13 Mechanisms proposed to explain the immunoprotective role of liver grafts include the induction of apoptosis of recipient T cells within the graft60, 61 or in recipient lymphoid organs.2, 14 We propose that among transferred donor-derived leukocytes, hepatic NK cells in particular may be instrumental in inducing apoptosis of alloreactive T cells of the recipient. Indeed, NK cells have been shown to kill allogeneic MDC in vivo in experimental transplantation models.23, 24, 62 Upon experimental allogeneic bone marrow transplantation donor-derived NK cells protect against graft rejection by killing recipient T cells and antigen presenting cells.26 We propose that a similar mechanism may be effective after LTX. Unfortunately, with current techniques it is not possible to detect killing of allogeneic leukocytes by freshly isolated NK cells in vitro.26 To study NK allo-reactivity ex vivo, prolonged activation in the presence of appropriately mismatched allogeneic feeder cells is needed,63-66 or use of NK-cell clones.26, 67-69 Similarly, only in vitro activated, but not resting, NK cells are able to kill monocyte-derived immature DC in vitro.70 Demonstrating cytotoxicity of ex vivo cultured or cloned liver NK cells against allogeneic leukocyte targets does not answer the question whether donor derived liver NK cells can kill recipient leukocytes in vivo. This question can therefore only be answered in an appropriate experimental animal liver transplantation model, which is beyond the scope of the present study.
Taken together, our data show that LTX in humans results in transfer of potentially cytotoxic, mainly CD56bright, NK cells of donor origin into the recipient. In addition CD56dim NK cells from the donor mix with the recirculating pool of recipient NK cells for approximately 2 weeks after LTX. Upon encounter with the recipient immune system, either within the graft or in recipient tissues, these NK cells may inhibit the recipient's response to the liver graft by killing recipient MDC or T cells. By analyzing associations between rejection-free liver graft survival and KIR - KIR ligand mismatching in donor to recipient direction, the influence of donor NK cells on the recipient's alloresponse in human LTX may be elucidated.
We like to thank all the transplant surgeons for collecting the perfusates during the backtable LTX procedure, and Nicole M. van Besouw, from the department of Internal Medicine at the Erasmus MC, for providing peripheral blood samples from RTX patients.
- 9Kinetics of intragraft cytokine expression, cellular infiltration, and cell death in rejection of renal allografts compared with acceptance of liver allografts in a rat model: early activation and apoptosis is associated with liver graft acceptance. Transplantation 1998; 65: 1370-1377., , , , , , et al.
- 39Characterization of the proliferative response to mitogens by liver-associated lymphocytes. In: 8th International Symposium on Cells of the Hepatic Sinusoid; September 1-5, 1996; Bordeaux, France., , , , , .