Association of peripheral NK cell counts with Helios⁺IFN‐γ^– T_regs in patients with good long‐term renal allograft function

Summary

Little is known about a possible interaction of natural killer (NK) cells with regulatory T cells (T_reg) in long‐term stable kidney transplant recipients. Absolute counts of lymphocyte and T_reg subsets were studied in whole blood samples of 136 long‐term stable renal transplant recipients and 52 healthy controls using eight‐colour fluorescence flow cytometry. Patients were 1946 ± 2201 days (153–10 268 days) post‐transplant and showed a serum creatinine of 1·7 ± 0·7 mg/dl. Renal transplant recipients investigated > 1·5 years post‐transplant showed higher total NK cell counts than recipients studied < 1·5 years after transplantation (P = 0·006). High NK cells were associated with high glomerular filtration rate (P = 0·002) and low serum creatinine (P = 0·005). Interestingly, high NK cells were associated with high CD4⁺CD25⁺CD127^–forkhead box protein 3 (FoxP3⁺) T_reg that co‐express the phenotype Helios⁺interferon (IFN)‐γ^– and appear to have stable FoxP3 expression and originate from the thymus. Furthermore, high total NK cells were associated with T_reg that co‐express the phenotypes interleukin (IL)−10^–transforming growth factor (TGF)‐β⁺ (P = 0·013), CD183⁺CD62L^– (P = 0·003), CD183⁺CD62⁺(P = 0·001), CD183^–CD62L⁺ (P = 0·002), CD252^–CD152⁺ (P < 0·001), CD28⁺human leucocyte antigen D‐related (HLA‐DR^–) (P = 0·002), CD28⁺HLA‐DR⁺ (P < 0·001), CD95⁺CD178^– (P < 0·001) and CD279^–CD152⁺ (P < 0·001), suggesting that these activated T_reg home in peripheral tissues and suppress effector cells via TGF‐β and cytotoxic T lymphocyte‐associated protein 4 (CTLA‐4). The higher numbers of NK and T_reg cell counts in patients with long‐term good allograft function and the statistical association of these two lymphocyte subsets with each other suggest a direct or indirect (via DC) interaction of these cell subpopulations that contributes to good long‐term allograft acceptance. Moreover, we speculate that regulatory NK cells are formed late post‐transplant that are able to inhibit graft‐reactive effector cells.

Introduction

The role of natural killer (NK) cells in solid organ and bone marrow transplantation remains somewhat unclear 1, 2. NK cells are thought to represent cytotoxic effector cells involved in immune responses against virus, tumour cells and allografts 3, 4, 5. Additionally, immunoregulatory NK cells with phenotype and cytokine patterns different from cytotoxic effector NK cells have been described, and are thought to be involved in successful pregnancy 4, 6, 7, 8. Biomarkers associated with tolerance following kidney transplantation are dominated by increases in peripheral B cell numbers and the expression of B cell‐related genes in blood lymphocytes, as well as defects in B cell maturation and evidence of active suppression mediated by B cells in vitro, as reviewed by Newell and Turka 9. Sagoo et al. reported that tolerant renal transplant recipients displayed an expansion of peripheral blood B and NK lymphocytes, reduced activated CD4⁺ T cells, a lack of donor‐specific antibodies, donor‐specific hyporesponsiveness of CD4⁺ T cells and a high ratio of forkhead box P3 (FoxP3) to alpha‐1,2‐mannosidase gene expression 10. In addition, biomarkers associated with tolerance following liver transplantation are related predominantly to NK cells and γδ T cells in the blood and genes related to iron homeostasis within the graft 9. Tolerant recipients display an increased proportion of NK cells and a decreased proportion of Vδ2 γδ T cells in their peripheral blood 9. An increase in the proportion of CD4⁺CD25⁺CD127^–FoxP3⁺ T cells in the blood was also noted, albeit only 12 months following discontinuation of immunosuppression 9. Control of NK cell function by CD4⁺CD25⁺ regulatory T cells was observed in several studies 11. Toll‐like receptor (TLR)−9‐engaged plasmacytoid dendritic cells (DC) represent a critical stimulus for human NK cells, and CD4⁺ T helper cells (Th) cells and CD4⁺CD25^hi regulatory T cells (T_reg) play an important role in modulating human NK cell responses 12. Jukes et al. reported on a bystander activation of invariant NK (iNK) T cells that occurs during conventional T cell alloresponses 13. In the setting of haematopoietic stem cell transplantation, murine and human in‐vitro experiments support the assertion that T_reg antagonize NK responses 2. Some studies support an augmented NK cell response in the absence of T_reg, as reviewed by Verneris 2. Others reported that murine NK cells delay allograft rejection in lymphopenic hosts by down‐regulating the homeostatic proliferation of CD8⁺ T cells 14. In haemodialyzed and kidney transplanted patients, innate‐like and conventional T cell populations were shown to be equally compromised 15. Padroza‐Pacheco et al. reviewed the interaction between NK cells and T_reg for future perspectives of immunotherapy 16.

In the present study, several T_reg subsets were analysed 17, 18, 19, 20, 21. Signalling through interferon (IFN)‐γR and interleukin (IL)−12R, in combination with T cell receptor (TCR) engagement, induces strong expression of the transcription factor T‐bet, which drives the differentiation of conventional T cells to a Th1 lineage 22. Up‐regulation of the transcription factor Helios modulates and stabilizes FoxP3 expression 23, 24, 25, 26, 27, 28, 29, 30. Helios is expressed on the majority of thymus‐derived T_reg (tT_reg) 31; however, part of the tT_reg lack Helios expression 32. Peripherally induced T_reg (pT_reg) do not express Helios 31, but are able to up‐regulate Helios in combination with FoxP3 during cell activation 33, 34. Thymic tT_reg were shown to have stable FoxP3 expression in contrast to peripherally induced pT_reg presenting FoxP3 only transiently, with the risk of reprogramming from T_reg to Th1 effector cells 19, 35. McClymont et al. reported that the majority of in‐vitro‐induced IFN‐γ⁺ T_reg did not express Helios, suggesting that they were generated extra‐thymically 36. Alternatively, they might belong to a minority of Helios^– tT_reg expressing FoxP3, CD39, cytotoxic T lymphocyte‐associated protein 4 (CTLA‐4), CCL3 and IFN‐γ, as published by Himmel et al. 32. Further experiments showed that tT_reg can polarize towards IFN‐γ⁺ T cells in‐vitro by interleukin (IL)−12 conditioning, whereby they remain Helios⁺, suggesting that part of the thymic‐derived T_reg population exhibits plasticity in cytokine production and expresses a Th1‐like phenotype 36. As shown in the blood of healthy individuals, Helios⁺IFN‐γ⁺ T_reg co‐express TGF‐β but not IL‐10. Further analysis of T_reg phenotypes showed that T_reg co‐expressed granzyme B and perforin in‐addition, as well as Fas (CD95) and FasL (CD178), thereby affording the T_reg the capacity to induce lysis and apoptosis of target cells 37. Moreover, expression of CTLA‐4 (CD152) and CD40L (CD154) imply cell–cell contact‐dependent immunosuppression by these T_reg subsets. CXCR3 and CD62L expression suggests that part of these cells have the potential to enter secondary lymphoid organs as well as inflamed tissues 38, 39. These T_reg exhibit Th1 characteristic properties such as IFN‐γR1 (CD119) and T‐bet expression, which means they have the potency to regulate expression as well as consumption of IFN‐γ in the cell. CD28 is involved in T_reg activation and human leucocyte antigen D‐related (HLA‐DR) expression indicates activation of T_reg 40.

A possible relationship or interaction of NK cells and T_reg in renal transplant recipients has not been examined previously. In the present study, we looked for a possible association of NK cells with certain T_reg subsets in patients with good long‐term renal allograft acceptance. If evidence for such an association could be found, it would suggest a direct or indirect (via DC) immunoregulatory interaction of these two lymphocyte subpopulations.

Methods

Healthy controls and patients

Laboratory staff served as healthy controls. All controls (n = 52) and renal transplant recipients (n = 136) gave informed consent for the tests performed within this study and the study was approved by the local ethical committee (S‐225/2014). The study was conducted in adherence to the Declaration of Helsinki. The mean age of patients was 49 ± 13 years (range = 19–76 years) and that of healthy controls was 41 ± 12 (range = 20–60) years (Table 1). Patients were 1946 ± 2201 days (range = 153–10 268 days) post‐transplant and showed a glomerular filtration rate (GFR) of 48 ± 20 ml/min (range = 11–111 ml/min) and a serum creatinine of 1·7 ± 0·7 mg/dl (range = 0·7–4·6 mg/dl) (Fig. 1). Blood for the tests was obtained during regular routine investigations in the out‐patient clinic. At the time of testing, the patients showed no signs of acute rejection or infection. All patients received standard immunosuppression, consisting of combinations of cyclosporin A, tacrolimus, steroids, mycophenolate mofetil or azathioprine (Table 1).

Table 1.

Demographic data of patients and healthy controls

Category	Patients, n (%)	Healthy Controls
Sex
Male	91 (67%)	26 (49%)
Female	45 (33%)	26 (51%)
Age, mean, years	49 ± 13	41 ± 12
Range, years	19 + 76	20 + 60
Graft no.
First	115 (84%)	+
Second	16 (12%),	+
Third	5 (4%)	+
End‐stage renal disease
Chronic GN	35 (26%)	+
Diabetes	8 (6%)	+
Hypertension/ischaemic	4 (3%)	+
Nephrosclerosis	8 (6%)	+
Pyelonephritis	8 (6%)	+
Polycystic	24 (18%)	+
IGA nephropathy	17 (13%)	+
Other	32 (24%)	+
Clinical follow up
5 months–1·5 years	39 (29%)	+
>1·5–5	54 (40%)	+
>5–10	19 (14%)	+
>10–20	14 (10%)	+
>20	10 (7%)	+
Donor
Living	54 (40%)	+
Deceased	82 (60%)	+
HLA‐ABDR MM
0	19 (14%)	+
1	6 (4%)	+
2	29 (21%)	+
3	34 (25%)	+
4	20 (15%)	+
5	24 (18%)	+
6	4 (3%)	+
ABO incompatible transplants
Yes	14 (10%)	+
No	122 (90%)	+
Serum creatinine, mg/dl
< 2	97 (71%)	+
≥ 2	39 (29%)	+
GFR, ml/min
< 60	101 (74%)	+
≥ 60	35 (26%)	+
Patients with
1st investigation	136 (100%)	+
2nd investigation	59 (43%)	+
3rd investigation	11 (8%)	+
Rejection
Banff Ia, Ib, II	9 (7%)	+
Non‐rejected	109 (80%)	+
Unknown	18 (13%)	+
Immunosuppressive protocol
Steroids	82 (60%)	+
Cyclosporin	20 (15%)	+
Tacrolimus	104 (76%)	+
MMF	91 (67%)	+
AZA	15 (11%)	+

GN = glomerulonephritis; IgA = immunoglobulin A; HLA = human leucocyte antigen; GFR = glomerular filtration rate; MMF = mycophenolate mofetil; AZA = azathioprine.

Figure 1.

Graft function of long‐term stable kidney transplant recipients. Patients (n = 136) were 1946 ± 2201 days (153–10 268 days) post‐transplant and showed a serum creatinine of 1·7 ± 0·7 mg/dl (0·7–4·6 mg/dl).

Determination of lymphocyte subpopulations

Because the combination of phycoerythrin‐labelled CD16 and CD56 monoclonal antibody was used, peripheral blood levels of total NK cells included CD16⁺CD56⁺, CD16^–CD56⁺ and CD16⁺CD56^– subpopulations of NK cells 18. In detail, using four‐colour fluorescence flow cytometry CD45⁺CD3⁺CD16^–CD56^–CD19^–, CD45⁺CD3^–CD16⁺/^–CD56⁺/^–CD19^–, CD45⁺CD3^–CD16^–CD56^–CD19⁺, CD45⁺CD3⁺CD4⁺CD8^– and CD45⁺CD3⁺CD4^–CD8⁺ lymphocyte subsets were defined. As isotype controls served immunoglobulin (Ig)G2a/fluorescein isothiocyanate, IgG2a/phycoerythrin, IgG2a/allophycocyanin and IgG2a/peridinin–chlorophyll–protein complex antibodies. All antibodies (CD45: clone 2D1; CD3: clone SK7; CD4: clone SK3, CD8: clone SK1; CD19: clone SJ25C1; CD16: clone B73.1; CD56: clone NCAM16.2) were purchased from Becton Dickinson (BD)/Pharmingen (Heidelberg, Germany). Ten microlitres (µl) of a mixture of four different monoclonal antibodies conjugated with fluorescein isothiocyanate, phycoerythrin, allophycocyanin or peridinin–chlorophyll–protein complex were added to 50 µl of heparinized whole blood and incubated for 15 min at room temperature. Erythrocytes were lysed with NH₄Cl for 15 min. Lymphocyte subsets were analysed using a fluorescence‐activated cell sorter (FACS)Calibur flow cytometer (BD Biosciences, Heidelberg, Germany). The flow cytometer was calibrated every day using CaliBRITE beads (BD Pharmingen) to ensure optimal counting.

Determination of different T_reg subsets

T_reg subsets were determined as described previously 21. The exact gate settings of the present study are depicted in Fig. 2. For analysis of determinants on the cell surface, peripheral blood leucocytes (PBL) were incubated with fluorochrome‐labelled monoclonal antibodies against CD4 (clone PRA‐4), CD25 (IL‐2R; clone M‐A251), CD28 (clone CD28‐2), CD62L (L‐selectin; clone DREG‐56), CD95 (Fas; clone DX2), CD119 (IFN‐γR; clone GIR‐208), CD127 (clone: HIL‐7R‐M21), CD152 (CTLA‐4; clone: BNI3), CD154 (CD40L; clone: TRAP1), CD178 (FasL; NOK‐1), CD252 (OX40L; clone: IK‐1), CD279 (PD‐1; clone MIH4), HLA‐DR (clone: G46‐6) and CD183 (CXCR3; clone IC6/CXCR3) (all from BD Biosciences). Intracellular determinants were stained with fluorochrome‐labelled monoclonal antibodies against FoxP3 (clone: 236A/E7), IFN‐γ (clone B27), IL‐10 (clone: JES3‐19F1), granzyme B (clone: GB11), perforin (clone: δG9), T‐bet (clone: 04‐46) (all from BD Biosciences), Helios (clone: 22F6) (eBioscience, Frankfurt, Germany) and TGF‐β₁ (clone: 27232) (R&D systems, Wiesbaden, Germany). Briefly, PBL were incubated with combinations of monoclonal antibodies for 30 min as described and eight‐colour fluorescence was analysed using a FACSCanto II triple‐laser flow cytometer (BD Biosciences). When, in addition, intracellular proteins were studied, cell membranes were permeabilized using BD perm/wash buffer (BD Biosciences). At least 100 000 events were analysed in the initial forward‐/side‐scatter (FSC/SSC) dot‐plot. Total CD4⁺CD25⁺FoxP3⁺CD127^– T_reg as well as subsets of CD4⁺CD25⁺FoxP3⁺ CD127^– T_reg co‐expressing CD28, CD62L (L‐selectin), CD95 (Fas), CD119 (IFN‐γR), CD152 (CTLA‐4), CD154 (CD40L), CD178 (FasL), CD252 (OX40L), CD279 [programmed death 1 (PD‐1)], HLA‐DR, CD183 (CXCR3), IFN‐γ, IL‐10, granzyme B, perforin, T‐bet, Helios and TGF‐β₁ were studied.

Figure 2.

Determination of regulatory T cell (T_reg) subsets. Gates were set according to the corresponding isotype controls (isotype controls not shown). Stepwise gating strategy for T_reg subset determination: first, lymphocytes gate (P1), then CD4⁺CD25⁺ peripheral blood lymphocyte (PBL) gate (P2), and finally forkhead box protein 3 (FoxP3⁺)CD127^– gate (P3). Further CD4⁺CD25⁺FoxP3⁺CD127^– T_reg subsets (based on gate P3) were analysed using the depicted gate settings for Helios/interferon (IFN)‐γ, interleukin (IL)−10/transforming growth factor (TGF)‐β, CD183/CD62L, CD252/CD152 (CD152 intracellular), CD119/T‐bet, CD28/human leucocyte antigen D‐related (HLA‐DR), CD152/CD154 (both surfaces), perforin/granzyme B, CD95/CD178, and CD279/CD152 (CD152 intracellular). The depicted analysis is based on 100 000 events in the initial forward‐/side‐scatter (FSC/SSC) dot‐plot (P1).

Statistical analysis

For statistical analysis, PASW statistics program version 21 (IBM, Chicago, IL, USA), Spearman's rank correlation and Mann–Whitney U‐tests were used. P‐values < 0·01 after Bonferroni's correction were considered significant. Simple linear regression was conducted first to determine the influence of NK cells, CD8⁺ and CD4⁺ T and CD19⁺ B lymphocytes on different T_reg subsets in univariate analysis. Thereafter, and before running multivariate regression analysis, cell subset distributions were explored and log‐transformed when necessary to meet the criteria of this analysis. A backstep elimination model was used in which variables that produced P‐values < 0·05 were considered.

Results

Demographic data are summarized in Table 1 and the panels of lymphocyte and CD4⁺CD25⁺FoxP3⁺CD127^– T_reg subsets analysed are shown in Table 2.

Table 2.

T lymphocyte and regulatory T cell (T_reg) subset counts in the blood of healthy individuals and patients

	Healthy controls (n = 52) mean ± s.d.	Patients (n = 136) mean ± s.d.	P *
Total CD45+ lymphocytes/µl	1836 ± 496	1444 ± 817	< 0·001
%CD3+	73 ± 5·2	78 ± 11	< 0·001
CD3+/µl	1345 ± 401	1151 ± 718	0·001
%CD4+	45 ± 8·92	45 ± 13	0·611
CD4+/µl	821 ± 298	666 ± 420	0·001
%CD8+	26 ± 7·4	31 ± 13	0·031
CD8+/µl	473 ± 201	461 ± 403	0·045
% Total NK (CD16+/–CD56+/–)	13 ± 4·3	13 ± 9·6	0·152
Total NK/µl (CD16+/–CD56+/–)	228 ± 98	165 ± 134	< 0·001
%CD19+	12 ± 4·8	6·8 ± 5·2	< 0·001
CD19+/µl	223 ± 97	101 ± 106	< 0·001
CD4+CD25+FoxP3+CD127–/µl (Treg)	6·3 ± 3·7	3·6 ± 3·4	< 0·001
IFN‐γ–Helios+ Treg/µl	3·4 ± 2·5	1·6 ± 1·8	< 0·001
IFN‐γ–Helios– Treg/µl	3·2 ± 1·8	2·3 ± 2·4	< 0·001
IFN‐γ+Helios+ Treg/µl	0·19 ± 0·37	0·09 ± 0·18	0·007
IFN‐γ+Helios– Treg/µl	0·08 ± 0·18	0·09 ± 0·16	0·586
IL‐10+TGF‐β– Treg/µl	0·00 ± 0·00	0·01 ± 0·08	0·383
IL‐10*TGF‐β+ Treg/µl	0·00 ± 0·00	0·02 ± 0·20	0·165
IL‐10–TGF‐β+ Treg/µl	0·63 ± 1·0	0·68 ± 0·88	0·380
CD183+CD62L– Treg/µl	0·31 ± 0·40	0·27 ± 0·37	0·442
CD183+CD62L+ Treg/µl	2·1 ± 1·7	0·93 ± 1·1	< 0·001
CD183–CD62L+ Treg/µl	3·1 ± 2·0	1·9 ± 2·0	0·001
CD252+CD152– Treg/µl	0·07 ± 0·23	0·07 ± 0·42	0·359
CD252+CD152+ Treg/µl	0·02 ± 0·12	0·01 ± 0·04	0·633
CD252–CD152+ Treg/µl	2·1 ± 1·7	0·92 ± 1·3	< 0·001
Perforin+granzyme B– Treg/µl	0·04 ± 0·08	0·07 ± 0·21	0·756
Perforin+granzyme B+ Treg/µl	4·9 ± 11	4·3 ± 15	0·912
Perforin–granzyme B+ Treg/µl	2·2 ± 4·1	4·2 ± 11	0·106
CD28+HLA‐DR– Treg/µl	2·7 ± 1·7	1·8 ± 1·7	< 0·001
CD28+HLADR+ Treg/µl	2·6 ± 2·4	1·2 ± 1·4	< 0·001
CD28–HLA‐DR+ Treg/µl	0·12 ± 0·18	0·24 ± 0·52	0·591
CD95+CD178– Treg/µl	4·7 ± 3·2	2·6 ± 2·4	< 0·001
CD95+CD178+ Treg/µl	0·01 ± 0·02	0·00 ± 0·02	0·382
CD95–CD178+ Treg/µl	0·02 ± 0·05	0·01 ± 0·04	0·353
CD152+CD154– Treg/µl	0·04 ± 0·07	0·06 ± 0·30	0·041
CD152+CD154+ Treg/µl	0·00 ± 0·00	0·00 ± 0·02	0·383
CD152–CD154+ Treg/µl	0·03 ± 0·10	0·00 ± 0·02	0·006
CD119+T‐bet– Treg/µl	0·21 ± 0·41	0·20 ± 0·26	0·816
CD119+T‐bet+ Treg/µl	0·01 ± 0·04	0·00 ± 0·03	0·512
CD119–T‐bet+ Treg/µl	0·03 ± 0·07	0·02 ± 0·06	0·210
CD279+CD152– Treg/µl	0·05 ± 0·19	0·06 ± 0·32	0·939
CD279+CD152+ Treg/µl	0·07 ± 0·14	0·02 ± 0·06	< 0·001
CD279–CD152+ Treg/µl	2·0 ± 1·5	0·84 ± 1·1	0·001

*Mann–Whitney U ‐test: patients versus healthy individuals. All data are given as mean ± standard deviation (s.d.). P‐values < 0·01 were considered significant (Bonferroni correction). HLA‐DR = human leucocyte antigen D‐related; FoxP3 = forkhead box protein 3; IL= interleukin; IFN = interferon; TGF = transforming growth factor; NK = natural killer.

Lymphocyte and T_reg subsets in patients and healthy controls

Compared to healthy controls, the patients with well‐functioning grafts showed lower relative and absolute counts of CD19⁺ B lymphocytes (both P < 0·001; Mann–Whitney U‐test), indicating a B lymphocyte deficiency in patients. Moreover, the patients showed decreased absolute counts of total CD45⁺ lymphocytes and total NK cells (both P < 0·001) and CD3⁺ and CD4⁺ T lymphocytes (both P = 0·001), but similar absolute counts of CD8⁺ T effector cells (P = 0·045) (Table 2). The data suggest that CD8⁺ T lymphocytes are decreased only marginally late post‐transplant in patients, whereas other T, B and NK cell subsets are reduced significantly compared to healthy individuals.

Compared to healthy controls, patients showed significantly decreased absolute counts of total CD4⁺CD25⁺FoxP3⁺CD127^– T_reg (P < 0·001), IFN‐γ^–Helios⁺ P < 0·001), IFN‐γ^–Helios^– (P < 0·001) and IFN‐γ⁺Helios⁺ T_reg (P = 0·007), whereas IFN‐γ⁺Helios^– T_reg were similar in patients and controls (P = 0·586). When additional subsets of CD4⁺CD25⁺FoxP3⁺CD127^– T_reg were analysed, patients showed lower CD183⁺CD62L⁺, CD183^–CD62L⁺, CD252^–CD152⁺, CD28⁺HLA‐DR^–, CD28⁺HLA‐DR⁺, CD95⁺CD178^– (all P < 0·001), CD152^–CD154⁺ (P = 0·006), CD279⁺CD152⁺ (P < 0·001) and CD279^–CD152⁺ T_reg (P = 0·001) than healthy controls (Table 2). It appears that activated T_reg with homing receptors for lymph nodes and receptors for B lymphocyte and DC interaction are reduced in patients, whereas other T_reg subsets expressing IL‐10, TGF‐β, CD178, CD119, T‐bet, granzyme and/or perforin are present at similar levels as those found in healthy controls (P > 0·01) (Table 2).

Stability of lymphocyte and T_reg subsets in patients

To study the stability of lymphocyte counts and T_reg phenotypes, lymphocyte and T_reg subsets were studied twice in 59 patients [interval first versus second investigation: 120 ± 47 days; mean ± standard deviation (s.d.)] and three times in 11 patients (interval second versus third investigation: 106 ± 19 days). Only CD95+CD178^– (first versus second investigation: 2·6 ± 2·3/µl versus 1·9 ± 1·7/µl; P = 0·004; Wilcoxon's signed rank test) and perforin⁺granzyme B⁺ (second versus third investigation: 3·0 ± 5·3/µl versus 4·1 ± 6·7/µl; P = 0·008) absolute T_reg numbers differed among the measurements (data not shown).

NK cells and time post‐transplant

When lymphocyte and T_reg subsets in patients ≤ 1·5 versus > 1·5 years post‐transplant (n = 39 versus n = 97) were compared, only total NK cells were higher in patients > 1·5 years post‐transplant (124 ± 107/µl versus 181 ± 140/µl; P = 0·006; Mann–Whitney U‐test), whereas the other lymphocyte and T_reg subsets were similar (data not shown). Patients ≤ 1·5 years post‐transplant were analysed after a mean follow‐up of 13 ± 3·1 months post‐transplant (range = 5–18 months) and patients >1·5 years post‐transplant after a follow‐up of 90 ± 83 months (range = 20–360 months) (Table 1).

Graft function and lymphocyte/T_reg counts

Patients had a mean GFR of 48 ± 20 ml/min (± s.d.) and a mean serum creatinine of 1·7 ± 0·71 mg/dl. Both higher GFR and lower serum creatinine were associated with higher total NK and CD19⁺ B cells as well as higher IFN‐γ^–Helios⁺, CD28⁺HLA‐DR⁺, CD252^–CD152⁺ and CD183⁺CD62L⁺ T_reg (all P < 0·01, Spearman's rank correlation test) (Table 3). GFR and serum creatinine were not associated with CD8⁺ lymphocyte counts (P = 0·147; P = 0·222).

Table 3.

Association of glomerular filtration rate (GFR) with lymphocyte and regulatory T cell (T_reg) subsets in 136 transplant recipients

	GFR		Creatinine
Parameter	P	r	P	r
Total NK/µl (CD16+/–56+/–)	0·002	0·263	0·005	−0·241
CD19+/µl	< 0·001	0·349	0·001	−0·281
IFN‐γ–Helios+ Treg/µl	< 0·001	0·300	< 0·001	−0·305
CD28+HLA‐DR+ Treg/µl	0·005	0·242	0·004	−0·249
CD252–CD152+ Treg/µl	0·007	0·233	0·004	−0·246
CD183+CD62L+ Treg/µl	< 0·001	0·305	< 0·001	−0·343

All parameters of Table 2 were analysed using Spearman's rank correlation test. Only parameters showing P‐values of P < 0·01 with both GFR and serum creatinine are shown in Table 3. HLA‐DR = human leucocyte antigen D‐related; NK = natural killer.

NK cells, CD8⁺ lymphocytes and total T_regs

High absolute numbers of total NK cells (P < 0·001) as well as high absolute counts of CD8⁺ lymphocytes (P = 0.011) were associated with CD4⁺CD25⁺CD127^–FoxP3⁺ T_reg counts (Table 4, Fig. 3). Healthy individuals lack this association. In backstep multivariate regression analysis with adjustment for the main variables CD8⁺ T lymphocytes and NK cells, in addition to CD19⁺ B and CD4⁺ T lymphocytes, we found that the positive association of NK cells with CD4⁺CD25⁺CD127^–FoxP3⁺ T_reg still existed (beta‐coefficient = 0·23, confidence interval (CI) = 0·07–0·4, P = 0·006), whereas the significant association between CD8⁺ T lymphocytes and T_reg was not maintained.

Table 4.

Associations of lymphocyte and regulatory T cell (T_reg) subsets in stable long‐term renal transplant recipients and healthy controls

	Healthy controls				Patients
	(n = 52)				(n = 136)
	CD8+/µl		Total NK/µl		CD8+/µl		Total NK/µl
Subset/µl	r	P	r	P	r	P	r	P
CD45+	0·554	< 0·001	0·433	0·001	0·749	< 0·001	0·411	< 0·001
CD3+	0·624	< 0·001	0·318	0·022	0·791	< 0·001	0·223	0·009
CD4+	0·167	0·235	0·253	0·070	0·483	< 0·001	0·285	0·001
CD8+			0·207	0·141			0·075	0·383
Total NK (CD16+/–CD56+/–)	0·207	0·141			0·075	0·383
CD19+	−0·025	0·859	−0·035	0·804	0·403	< 0·001	0·213	0·013
Treg (CD4+CD25+CD127–FoxP3+)	−0·202	0·155	−0·029	0·842	0·219	0·011	0·347	< 0·001
Helios+IFN‐γ– Treg	−0·249	0·078	−0·096	0·503	0·065	0·453	0·339	< 0·001
Helios+IFN‐γ+ Treg	−0·198	0·164	−0·184	0·197	0·014	0·870	0·033	0·703
Helios–IFN‐γ+ Treg	−0·344	0·014	−0·008	0·955	−0·041	0·633	0·055	0·524
IL‐10–TGF‐β+ Treg	−0·109	0·445	−0·089	0·536	0·215	0·012	0·212	0·013
CD183+CD62L– Treg	−0·110	0·440	0·172	0·226	0·181	0·036	0·254	0·003
CD183+CD62L+ Treg	−0·115	0·422	0·107	0·456	0·164	0·057	0·295	0·001
CD183–CD62L+ Treg	0·019	0·893	−0·037	0·794	0·188	0·029	0·263	0·002
CD252+CD152– Treg	−0·193	0·174	−0·225	0·112	0·053	0·542	0·068	0·433
CD252+CD152+ Treg	−0·306	0·029	−0·280	0·047	0·166	0·054	0·040	0·645
CD252–CD152+ Treg	−0·224	0·114	−0·006	0·969	0·134	0·123	0·421	< 0·001
Perforin+granzyme B– Treg	−0·030	0·836	0·116	0·419	−0·053	0·543	0·073	0·399
Perforin+granzyme B+ Treg	0·054	0·707	0·175	0·219	0·383	< 0·001	−0·115	0·183
Perforin– granzyme B+ Treg	0·331	0·018	−0·027	0·851	0·515	< 0·001	−0·186	0·031
CD28+HLA‐DR– Treg	−0·095	0·508	−0·095	0·508	0·181	0·036	0·270	0·002
CD28+HLA‐DR+ Treg	−0·307	0·028	−0·012	0·931	0·122	0·160	0·407	< 0·001
CD28–HLA‐DR+ Treg	0·140	0·326	0·235	0·097	0·272	0·001	0·073	0·399
CD95+CD178– Treg	−0·248	0·079	−0·173	0·226	0·116	0·181	0·359	< 0·001
CD95+CD178+ Treg	−0·066	0·647	0·069	0·630	0·049	0·575	0·023	0·788
CD95–CD178+ Treg	−0·121	0·397	−0·134	0·349	−0·284	0·001	0·008	0·926
CD152+CD154– Treg	−0·044	0·757	0·148	0·299	0·019	0·829	−0·006	0·943
CD152+CD154+ Treg					0·006	0·948	−0·149	0·084
CD152–CD154+ Treg	0·013	0·926	−0·103	0·471	−0·083	0·341	0·031	0·722
CD119+T‐bet– Treg	−0·004	0·979	0·232	0·102	0·156	0·070	0·104	0·231
CD119+T‐bet+ Treg	−0·071	0·619	0·054	0·705	0·052	0·553	0·144	0·097
CD119–T‐bet+ Treg	0·120	0·402	−0·072	0·616	0·115	0·182	−0·093	0·284
CD279+CD152– Treg	−0·041	0·773	−0·035	0·810	−0·023	0·789	−0·023	0·787
CD279+CD152+ Treg	0·027	0·853	0·359	0·010	0·112	0·197	0·148	0·087
CD279–CD152+ Treg	−0·256	0·069	−0·078	0·587	0·171	0·048	0·372	< 0·001

Statistical analysis was performed using Spearman's rank correlation test. P‐values of <0·02 were marked by colours and correlations of CD8⁺/µl were indicated by orange and correlations of total natural killer (NK) cells/µl by yellow background. HLA‐DR = human leucocyte antigen D‐related; FoxP3 = forkhead box protein 3; IL= interleukin; IFN = interferon; TGF = transforming growth factor.

Figure 3.

Associations of total natural killer (NK) cell counts with graft outcome and regulatory T cell (T_reg) subset counts in long‐term stable renal transplant recipients. Absolute counts of total NK cells were associated with glomerular filtration rate (r = 0·263; P = 0·002; Spearman's rank test) and serum creatinine (r = –0·241; P = 0·005) as well as absolute total T_reg (r = 0·347; P < 0·001) and Helios⁺interferon (IFN)‐γ^– T_reg counts (r = 0·339; P < 0·001) in long‐term stable renal transplant recipients (n = 136).

NK cells, CD8⁺ lymphocytes and Helios⁺IFN‐γ^– T_regs

Interestingly, absolute numbers of total NK cells were found to increase in association with Helios⁺IFN‐γ^– T_reg (P < 0·001), whereas CD8⁺ lymphocytes lacked this relationship with Helios⁺IFN‐γ^– T_reg (P = 0·453) (Fig. 3, Table 4). In backstep multivariate regression analysis with adjustment for CD8⁺ T lymphocytes and NK cells as the main variables, in addition to CD19⁺ B and CD4⁺ T lymphocytes, we found that the positive association of NK cells with Helios⁺IFN‐γ^– T_reg still existed (beta‐coefficient = 0·38, CI = 0·18–0·6, P < 0·001).

NK cells, CD8⁺ lymphocytes and additional T_reg subsets

Total NK cells were associated with CD4⁺CD25⁺CD127^–FoxP3⁺ T_reg that co‐express the phenotypes IL‐10^–TGF‐β⁺ (P = 0·013), CD183⁺CD62L^– (P = 0·003), CD183⁺CD62L⁺ (P = 0·001), CD183^–CD62L⁺ (P = 0·002), CD252^–CD152⁺ (P < 0·001), CD28⁺HLA‐DR^– (P = 0·002), CD28⁺HLA‐DR⁺ (P < 0·001), CD95⁺CD178^– (P < 0·001) and CD279^–CD152⁺ (P < 0·001) (Table 4). In backstep multivariate regression analyses with adjustment for the variables CD8⁺ T lymphocytes and NK cells as the main variables, in addition to CD19⁺ B and CD4⁺ T lymphocytes, we found that the positive association of NK cells with only CD183⁺CD62L^– (beta‐coefficient = 0·38, P < 0·001), CD252^–CD152⁺ (beta‐coefficient = 0·56, P < 0·001), CD28⁺HLA‐DR⁺ (beta‐coefficient = 0·28, P = 0·011), CD95⁺CD178^– (beta‐coefficient = 0·27, P = 0·002) and CD279^–CD152⁺ T_reg (beta‐coefficient = 0·27, P = 0·002) still existed.

Notably, absolute numbers of CD8⁺ T lymphocytes were associated with perforin⁺granzyme B⁺ (P < 0·001), perforin^–granzyme B⁺ (P < 0·001), CD95^–CD178⁺ (P = 0·001) and CD28^–HLA‐DR⁺ (P = 0·001) T_reg, whereas total NK cells did not show these associations (P = 0·183; P = 0·031; P = 0·926; and P = 0·399, respectively) (Table 4). In backstep multivariate regression analysis with adjustment for the variables NK, CD19⁺ B and CD4⁺ T lymphocytes, we found that the positive association of CD8⁺ T lymphocytes with perforin^–granzyme B⁺ T_reg remained significant (beta‐coefficient = 0·91, P < 0·001).

Healthy controls showed an association of absolute NK cell counts only with CD279⁺CD152⁺ T_reg (P = 0·010) (Table 4). In healthy individuals, high CD8⁺ lymphocytes were associated with low Helios^–IFN‐γ⁺ (P = 0·014) and high perforin^–granzyme B⁺ (P = 0·018) T_reg.

NK cells, CD8⁺ lymphocytes and lymphocyte subsets

Interestingly, an association of CD8⁺ lymphocytes as well as total NK cells with CD19⁺ B lymphocytes was only found in patients (P < 0·001; P = 0·013) and so was an association with CD4⁺ T lymphocyte counts (P < 0·001; P = 0·001, respectively). In backstep multivariate regression analysis with adjustment for the variables NK, T_reg and CD4⁺ T lymphocytes, we found that the significant positive association of CD8⁺ T lymphocytes or NK cells with CD19⁺ B lymphocytes was not maintained. Conversely, the significant association of CD4⁺ T lymphocytes with CD8⁺ T lymphocytes remained significant in multivariate analysis (beta‐coefficient = 0·33, P < 0·001). In healthy individuals these associations were absent (P = 0·859; P = 0·804; P = 0·235; and P = 0·070, respectively) (Table 4). The data suggest that CD8⁺ T lymphocyte counts are associated independently with CD4⁺ T lymphocyte counts in immunosuppressed patients with well‐functioning grafts.

Immunosuppressive protocol and lymphocyte/T_reg counts

We studied whether treatment with certain immunosuppressive drugs was associated with certain lymphocyte and T_reg counts. Patients with higher daily steroid doses also received higher azathioprine doses (r = 0·806, P = 0·009; Spearman's rank correlation test), and patients with higher tacrolimus doses tended to receive, in addition, higher mycophenolate doses (r = 0·343, P = 0·003). Interestingly, steroids (n = 82), cyclosporin (n = 20), tacrolimus (n = 104) and mycophenolate mofetil (n = 91) doses did not show significant associations with absolute numbers of T and B lymphocytes, NK cells or T_reg subsets. In contrast, high azathioprine doses were associated with low absolute counts of total NK cells (n = 15; r = –0·775, P < 0·001). Interestingly, when patients with or without steroid therapy were compared, patients on steroids had higher CD28⁺HLA‐DR^– (P = 0·005) and lower CD28^–HLA‐DR⁺ (P = 0·002) and CD279⁺CD152^– T_reg (P = 0·008) counts than steroid‐free patients (n = 49)· The other cell subsets were similar in patients on steroid‐containing or steroid‐free immunosuppressive maintenance. Blood levels of cyclosporin (n = 20 patients), tacrolimus (n = 104) and mycophenolate mofetil (n = 44) were not associated significantly with lymphocyte and T_reg subset counts.

Immunosuppressive protocol and GFR

There was no association of GFR with daily oral drug dose or blood level of immunosuppressive drugs.

Immunosuppressive drugs, lymphocyte and T_reg subsets and time post‐transplant

Daily mycophenolate mofetil (r = –0·319; P = 0·002; Spearman's rank correlation test) and tacrolimus doses (r = –0·290; P = 0·003) as well as mycophenolate blood levels (r = –0·486; P = 0·001) were decreased with time post‐transplant, and CD28^–HLA‐DR⁺ T_reg counts (r = 0·300; P < 0·001) increased simultaneously.

Rejection and lymphocyte/T_reg subsets

Ninety‐three patients never experienced rejection, nine had experienced a rejection Banff grade > 1 prior to testing, and 16 had experienced previous borderline rejection. All rejections were reversed and all patients had well‐functioning kidney grafts at the time of testing. Lymphocyte and T_reg subsets were not significantly different in the three patient groups.

Infection and lymphocyte/T_reg subsets

Patients were free of acute infection at the time of testing. Six patients had experienced viral infections and 25 bacterial infections more than 3 months prior to testing. Previous viral infection was not associated with lymphocyte and T_reg subsets (P > 0·01; Mann–Whitney U‐test), whereas previous bacterial infection was associated with increased IL‐10⁺TGF‐β⁺ T_reg (P < 0·001).

Discussion

In the present study, we investigated whether there is an association of NK cells with T_reg subsets in peripheral blood in kidney recipients with good long‐term allograft outcome. Peripheral blood levels of total NK cells included CD16⁺CD56⁺, CD16^–CD56⁺ and CD16⁺CD56^– subpopulations of NK cells. Our data suggest that NK cell counts increase with time post‐transplant in patients with good graft outcome. When lymphocyte and T_reg subsets in patients ≤ 1·5 versus > 1·5 years post‐transplant were compared, only total NK cells were higher in patients > 1·5 years post‐transplant, whereas the other lymphocyte and T_reg subsets were similar. The long‐term NK cell increase was associated neither with daily doses of immunosuppressive drugs or blood levels of immunosuppressants nor with the occurrence of acute infection or rejection. We found evidence to suggest that NK cell counts increase independently in parallel with T_reg counts, particularly Helios⁺IFN‐γ^– thymus‐derived tT_reg. This particular T_reg subset co‐expresses the activation marker HLA‐DR and appears to affect effector cells functionally by release of TGF‐β or via CTLA‐4‐mediated cell interaction with DC in lymph nodes. These associations were observed in transplant patients, but not in healthy individuals. We therefore speculate that whereas healthy controls have stable NK cells counts, NK cell and T_reg counts increase with time post‐transplant in patients with good graft function and direct or indirect (via DC) interaction of these cell subsets may prevent graft damage by the innate immune system. The stimulus for the NK cell increase remains unknown. Interestingly, CD8⁺ lymphocytes did not show a similar increase post‐transplant; these cells were associated strongly with activated HLA‐DR expressing T_reg that co‐express apoptosis‐inducing substances and determinants such as perforin, granzyme B and Fas ligand. One might speculate that graft‐specific CD8⁺ lymphocytes were killed by cytotoxic T_reg, thereby preventing increases of CD8⁺ effector cells and keeping post‐transplant CD8⁺ lymphocyte counts at a stable level. Stable levels of CD8⁺ effector cells were observed together with a lack of association of CD8⁺ lymphocytes with graft function, such as GFR and serum creatinine. Both these indicators of graft function were associated with NK cell counts; namely, high NK cells post‐transplant were associated with increased GFR and decreased serum creatinine. In other words, the data show that high NK cells are not harmful for the graft and instead are associated with good long‐term graft function. Because of the association of NK cell and T_reg counts we speculate that high NK cells may play a causative role in relation to high T_reg counts, and that T_reg may inhibit NK cell function via an as yet‐unknown pathway. Several pathways of NK cell inhibition have been described in animal and cell culture experiments and in clinical haematopoietic stem cell transplantation, and these observations are compatible with our findings in renal transplant recipients.

TGF‐β‐mediated suppression of NK cell function by T_reg has been observed in mice by Barao et al. 41, who demonstrated that the prior removal of host T_reg cells, but not CD8⁺ T cells, enhanced NK cell‐mediated bone marrow rejection significantly. The inhibitory role of T_reg on NK cells was confirmed in vivo with adoptive transfer studies in which transferred CD4⁺CD25⁺ cells could abrogate NK cell‐mediated hybrid resistance. Anti‐TGF‐β monoclonal antibody treatment also increased NK cell‐mediated bone marrow graft rejection, suggesting that the NK cell suppression was mediated by TGF‐β. The authors concluded that CD4⁺CD25⁺ T_reg cells can inhibit NK cell function in vivo potently, and that their depletion may have therapeutic ramifications for NK cell function in bone marrow transplantation and cancer therapy 41. TGF‐β‐dependent inhibition of NK cells by T_reg was also described by Ghiringhelli 42. Zhou et al. observed that in‐vitro‐induced T_reg cells decreased NK cell cytotoxicity and down‐regulated dramatically the IFN‐γ secretion of NK cells responding to IL‐12 stimulation 43. Moreover, it was found that cell–cell interaction was essential for suppression of NK cell function, and that TGF‐β played a vital role in the inhibition process. These results suggest that FoxP3 expression in polyclonal CD4⁺ T cells can induce T_reg cells and potentially inhibit NK cell function 43. Our finding of a strong association of TGF‐β‐producing T_reg with NK cell numbers suggests that T_reg might inhibit NK cells in renal transplant recipients directly by release of TGF‐β.

Others reported that NK T cells promoted changes in expression of negative co‐stimulatory receptors and anti‐inflammatory cytokines by T_reg and other T cell subsets in an IL‐4‐dependent manner, an effect that resulted in tolerance to bone marrow and organ grafts in animals 44. Additionally, NK cells are involved both in rejection and tolerance of solid allografts, as reviewed by Benichou et al. 45. Allo‐NKs were reported to be capable of inducing systemic tolerance after haplo‐ haematopoietic stem cell transplantation (HSCT) by assembling donor‐derived immature DCs to expand recipient‐derived T_reg cells in the thymus 46. Our finding of a strong association of CD152⁺ T_reg with NK cell numbers in renal transplant recipients suggests that T_reg might inhibit NK cells via negative signalling to DC that (a) might inhibit NK cells by release of immunosuppressive cytokines or negative signals during cell–cell contact, or (b) might induce further T_reg.

In kidney transplant patients, down‐modulation of CD16 and CD6 on CD56^(dim) NK cells was observed, with significant differences between cyclosporin A‐ and tacrolimus‐treated patients 47. Tacrolimus treatment was associated with decreased CD69, HLA‐DR and increased CD94/NKG2A expression in CD56^(dim) NK cells, indicating that the quality of immunosuppressive treatment impinges upon the peripheral NK cell repertoire. In‐vitro studies with peripheral blood mononuclear cells of healthy donors showed that this modulation of CD16, CD6, CD69 and HLA‐DR could also be induced experimentally. The presence of calcineurin or mTOR inhibitors had also functional consequences regarding degranulation and IFN‐γ production against K562 target cells, respectively 47. Furthermore, Bergmann et al. concluded from in‐vitro experiments that human tumour iT_reg cells inhibited IL‐2‐mediated NK cell activity in the absence of target cells, whereas, surprisingly, the tumoricidal activity of NK cells was enhanced by iT_reg cells 48. Gasteiger et al. reported that CD127⁺ NK cells expanded in an IL‐2‐dependent manner upon T_reg cell depletion and were able to give rise to mature NK cells, indicating that the latter can develop through a CD25⁺ intermediate stage 49. Thus, T_reg cells restrain the IL‐2‐dependent CD4⁺ T cell help for CD127⁺ immature NK cells. These findings highlight the adaptive control of innate lymphocyte homeostasis 49. The same authors reported that IL‐2 rapidly boosted the capacity of NK cells to engage target cells productively and enabled NK cell responses to weak stimulation 50. The results suggest that IL‐2‐dependent adaptive‐innate lymphocyte cross‐talk tunes NK cell reactivity and that T_reg cells restrain NK cell cytotoxicity by limiting the availability of IL‐2 50. The IL‐2 dependence of T_reg/NK cell interaction was confirmed in a murine diabetes model. Results from gene signature analyses, quantification of signal tranducer and activator of transcription (STAT)−5 phosphorylation levels, cytokine neutralization experiments, cytokine supplementation studies and evaluations of intracellular cytokine levels argue collectively for a scenario in which T_reg regulate NK cell functions by controlling the bioavailability of limiting amounts of IL‐2 in the islets, generated mainly by infiltrating CD4⁺ T cells 51. Moreover, Jukes et al. reported that iNK T cells are not activated by allogeneic cells per se, but expand, both in‐vitro and in‐vivo, in the presence of a concomitant conventional T cell response to alloantigen that is triggered by IL‐2 13. Our finding of an association of DR⁺Helios IFN‐γ^– T_reg with NK cell numbers suggests that most of the T_reg might have stable FoxP3 expression, originate from the thymus and therefore do not produce IL‐2 by themselves, and that these DR⁺ T_reg are activated and consume IL‐2, resulting in IL‐2‐dependent inhibition of NK cells. Immunosuppressive drugs such as cyclosporin and tacrolimus that inhibit IL‐2 production in lymphocytes can therefore contribute indirectly to NK cell suppression in renal transplant recipients.

In general, the cited experiments in animals and cell cultures as well as the observations in clinical haematopoietic stem cell transplantation suggest that T_reg interact directly or via DC with NK cells, with the potential to down‐regulate cytotoxic or to enhance immunoregulatory NK cell function. These observations agree with our finding that high NK cell counts are associated with high CD152⁺ T_reg that might transform immunostimulatory DC to immunosuppressive NK cell‐inhibiting DC. TGF‐β‐producing T_reg might suppress NK cells directly. Strongly activated thymical‐derived T_reg might inhibit NK cells by IL‐2 consumption. Our data suggest that NK cells might interact with T_reg in lymph nodes.

Moreover, we speculate that renal transplant recipients > 1·5 years post‐transplant develop an immunoregulatory NK cell type with diminished natural cytotoxicity and little IFN‐γ production upon monokine stimulation 52, similar to those immunoregulatory NK cells in the uterus that protect pregnancy 4, 53. Immunoregulatory NK cells (NK_reg) have been demonstrated in vitro 54, 55, and these NK_reg were able to suppress T cell responses antigen‐specifically in cell culture experiments 55. The hypothesis that late post‐transplant NK_reg in combination with T_reg and B_reg are generated is in line with our finding of high NK, B and T_reg cells in patients with good long‐term graft function and low immunosuppression, and is supported by reports of high NK, B and T_reg cells in tolerant kidney and liver transplant recipients 9, 10, 56, 57. Our data suggest that induction of NK, B and T_reg cells is a phenomenon late post‐transplant that is associated with good long‐term graft outcome and could be interpreted as an induction of NK, B and T lymphocytes with immunoregulatory function that might inhibit NK, B and T lymphocytes with effector function against the graft.

Conclusions

The higher numbers of peripheral NK cells late post‐transplant compared to early post‐transplant, the strong association of NK cells with T_reg and the strong association of graft function with high NK, T_reg and B cell counts in kidney recipients with good long‐term graft function suggest an interaction of these cell subpopulations that contributes to good long‐term allograft acceptance. Further in‐vitro studies are required to illuminate the mechanisms of NK/B/T_reg cell interaction in this particular patient cohort.

Limitations of the study

We studied only statistical associations showing good stable long‐term graft function in patients with high NK, B and T_reg cells. Further studies investigating phenotype and in‐vitro function of these lymphocyte subsets will clarify whether subsets with an immunoregulatory phenotype are induced, whether these subsets are able to interact with each other and whether they inhibit graft‐reactive effector cells.

Author contributions

V. D. designed the study and wrote the manuscript. G. O. made substantial contributions to conception and design as well as analysis and interpretation of data. K. T., L. Z., M. A., R. W., N. B. and C. M. have been involved in drafting the manuscript and revising it critically for important intellectual content. K. T. and M. A. were also involved in compilation and statistical analysis of the data. R. W., N. B. and C. M. treated the patients. All authors have given final approval of the version to be published.

Disclosure

The authors declare that they have no disclosures.