Terminally Differentiated Effector Memory CD8+ T Cells Identify Kidney Transplant Recipients at High Risk of Graft Failure : Journal of the American Society of Nephrology

Journal Logo

Clinical Research

Terminally Differentiated Effector Memory CD8+ T Cells Identify Kidney Transplant Recipients at High Risk of Graft Failure

Jacquemont, Lola1,2; Tilly, Gaëlle1,2; Yap, Michelle1,2; Doan-Ngoc, Tra-My1,2; Danger, Richard1,2; Guérif, Pierrick2; Delbos, Florent3; Martinet, Bernard1,2; Giral, Magali1,2; Foucher, Yohann4; Brouard, Sophie1,2; Degauque, Nicolas1,2

Author Information
JASN 31(4):p 876-891, April 2020. | DOI: 10.1681/ASN.2019080847
  • Free
  • Infographic
  • SDC

Abstract

Because kidney transplantation remains the best therapeutic option for ESKD, and considering that current immunosuppression approaches inefficiently treat chronic rejection, favorable kidney graft survival requires the design of innovative preventive tools and therapeutics adapted to patients’ individual risks. A better understanding of the immune response resulting from chronic allogeneic stimulation is thus needed (1) to identify novel biomarkers that anticipate the risk of allograft injury, and (2) to provide new perspectives on the identification of new therapeutic targets to prolong allograft survival.

The characterization of the humoral response has been heavily scrutinized over the last decades and has led to intensive development of monitoring tools that detect the presence of donor-specific anti-HLA antibodies and C1q binding activity as markers of kidney graft rejection.1–3 Quantification of anti-HLA donor-specific antibody (DSA) using single-antigen beads based on the mean fluorescence intensity clarified that the intrinsic properties of DSAs (e.g., C1q-, C3d-, and C4d-binding DSA IgG3 subclasses) are associated with an increase in kidney graft failure beyond the titer of DSA.1,4,5 For instance, Bouquegneau et al.6 reported an increase in allograft loss (hazard ratio [HR], 3.26) in patients with circulating complement-activating DSA compared with that in patients without complement-activating DSA in a large meta-analysis of 37 studies. In addition to complement-dependent mechanisms, the Fc segment of IgG may stimulate CD16-dependent cytotoxicity and inflammatory functions such as those of innate immune natural killer (NK) cells. An NK-cell molecular signature has been identified in biopsies from recipients of a kidney transplant (KTx) with antibody-mediated chronic rejection (ABMR).7–9 Histologic analysis of biopsies from KTx with C4d+ and C4d ABMR showed an enrichment of NK cell and macrophage infiltration according to CD56 and CD68 expression compared with that in KTx with T cell–mediated rejection.7 Interestingly, similar infiltration by CD3+ cells was found in biopsies from KTx diagnosed with ABMR or T cell–mediated rejection.7 Moreover, the NK-related gene signature differed between different studies,8–10 with an absence of CD16 in some studies7–9; interestingly, the genes used to define the NK signature (FGFBP2, SH21B, MYBL1, CX3CR1, KLRF1, GNLY, and CD16) were also overexpressed by effector memory (EM) T cells expressing CD45RA (terminally differentiated EM [TEMRA]) CD8 T cells (Supplemental Figure 1). This shared ABMR signature between NK and TEMRA CD8 cells prompted us to revisit the involvement of CD8 T cells in the process of kidney allograft rejection.

Although complement-dependent mechanisms were first considered the main pathway leading to the pathogenic effects of alloantibodies, the role of the cellular immune response is clearly not limited to the early post-transplant period but also participates in the late humoral and cytotoxic responses of chronic rejection. Memory T cells are considered a main obstacle to achieving transplantation success and even transplantation tolerance. Preexisting memory T cells are associated with high incidence rates of severe rejection episodes,11 and KTx prone to acute rejection have a higher precursor frequency of alloreactive CD8 T cells than patients who were nonrejectors.12 The involvement of CD8 T cells is not restricted to early post-transplant events. KTx with biopsy-proven ABMR exhibit high levels of the cytotoxic molecule granzyme B (GZMB)13–15 within their grafts and an increase in circulating CD8+CD28 T cells, with attrition of T cell receptor (TCR) variable β-chain (TCRVβ) repertoire diversity and high levels of IFN-γ, GZMB, and perforin (PERF-1).16–18 Interestingly, similar attrition of TCRVβ repertoire diversity was observed in KTx despite stable graft function for several years18 and was shown to be associated with an expansion of TEMRA (CD45RA+CCR7) CD8 T cells.18,19 Studies with clinical follow-up showed that KTx with a high frequency of TEMRA CD8 T cells exhibit a twofold higher risk of kidney dysfunction than those with a low frequency of TEMRA CD8 cells.19 However, the factors that regulate the expansion and function of TEMRA T cells, as well as their restriction toward donor antigens, remain poorly defined. We recently provided evidence that IL-15 is a potent activator of TEMRA CD8 cells from KTx and healthy volunteers (HV)20 and that, upon IL-15 stimulation, TEMRA CD8 cells from KTx promote inflammation by inducing the expression of inflammatory CX3CL1/fractalkine by endothelial cells in an IFN-γ– and TNF-α–dependent manner.20

Although risk factors of graft failure are well known, a challenging issue in kidney transplantation is to predict outcomes to help physicians guide patient care. In 2010, we created the Kidney Transplant Failure Score (KTFS),21 which is a composite score calculated 1-year post-transplantation using eight accepted clinical pre- and post-transplantation variables (recipient gender and age, donor creatinemia, number of previous transplantations, recipient creatinemia at 3 and 12 months, recipient proteinuria at 12 months, and occurrence of acute rejection within the first year). The KTFS is associated with an area under the time-dependent receiver operating characteristic (ROC) curve (AUC) of 0.78 for a prognostic of kidney graft survival up to 8 years post-transplantation, which is better than 1 year creatinemia or 1 year eGFR. The KTFS is presented as a methodologic model of predictive tools22 and is currently under clinical evaluation for its efficiency to drive personalized follow-up by video conferencing at home instead of face-to-face outpatient visits.23 In accordance with Braun et al.24 and Moore et al.,25 we believe that associating biologic—especially immunologic—biomarkers to such complex clinical predictors could enhance their clinical utility and improve personalization of patient follow-up.

In this study based on an independent cohort of 284 KTx, we showed that the measurement of CD8 memory subsets at 1 year post-transplant, in association with the clinical metrics summarized within the KTFS,21 allows improved identification of patients who will later return to dialysis as a consequence of graft failure. We demonstrated that donor-specific stimulation similarly activates a rapid and early response of TEMRA and EM (CD45RACCR7) CD8 T cells. Finally, we identified a unique innate-like signature of TEMRA CD8 T cells using RNA sequencing (RNA-seq) and mass cytometry (CyTOF) approaches and revealed a dual-trigger mechanism for TEMRA CD8 activation mediated by the engagement of TCR or CD16 pathways that results in the secretion of proinflammatory cytokines and the activation of the associated cytotoxic response.

Methods

Blood Samples

PBMCs were separated from blood samples collected in EDTA tubes on a Ficoll gradient layer according to the manufacturer’s recommendations and were frozen in DMSO supplemented with 10% autologous serum.

Polychromatic Flow Cytometry

The antibodies used for cytometric analyses are listed in Supplemental Table 1. A total of 2×106 frozen PBMCs were surface stained with antibodies specific for CD3, CD8, CD45RA, CCR7, CD27, CD28, and CD57. In addition to this cocktail, intracellular staining was performed using antibodies directed against T-bet, GZMB, PERF-1, and EOMES after fixation and permeabilization (Intracellular Fixation & Permeabilization Buffer Set; Thermo Fisher). A Yellow LIVE/DEAD Fixable Dead Cell Stain Kit was used to exclude dead cells from the analysis. BD CompBeads stained separately with individual mAbs were used to define the compensation matrix. Cells were analyzed with an LSRII flow cytometer (BD Immunocytometry Systems). Data were analyzed using FlowJo version 9.7.6 (TreeStar). For the assessment of CD16 expression by CD8 T cell subsets, PBMCs were stained with antibodies against CD3, CD8, CD45RA, CCR7, and CD16. CyTOF data from Bengsch et al.26 were analyzed to investigate the phenotype of CD16-expressing CD8 T lymphocytes. CD8 T lymphocytes were identified as iridium intercalator–positive, singlet LD–negative CD45+TCRγδCD3+CD4CD8α+ T cells and were then clustered according to the expression of 24 markers using Cytofkit and PhenoGraph.27–29 We then curated 23 high-dimensional clusters identified by PhenoGraph.

FACS Sorting, Cell Culture, and Multiplex Cytokine Production Measurement

Mixed Lymphocyte Reaction

PBMCs were thawed, rested overnight (O/N) in TexMacs Medium (Miltenyi), and stained for cell surface markers (CD3, CD8, CD45RA, and CCR7). Naive (CD45RA+CCR7+), EM (CD45RACCR7), and TEMRA (CD45RA+CCR7) CD3+CD8+ cells from KTx were FACS sorted (purity >95%, FACSAria; BD Biosciences). Donor PBMCs were T-cell depleted using a Pan T Cell Isolation Kit (Miltenyi) and irradiated at 35 Gy. T cell–depleted donor PBMCs were cocultured with CPDeFluor405 KTx cells at a 1:2 ratio in 96-well, round-bottom plates at 37°C in 5% carbon dioxide (CO2) in the presence of anti-CD107a mAb (except for proliferation assays). Cells were harvested and stained with anti-CD25 and anti-CD69 mAbs (day 1 and day 2) or with anti-CD3 mAb and Annexin V-FITC for proliferation assays (day 5). The production of 13 analytes was measured in culture supernatant from CD8 T cell subsets using a LEGENDplex Human CD8/NK Panel 13-plex (BioLegend).

CD16 Stimulation

PBMCs were thawed and rested O/N in TexMacs medium, and CD8 T cells were purified using a CD8 T Cell Isolation Kit. Purified CD8 T cells were cultured for 6 hours in 96-well, flat-bottomed plates at 37°C in 5% CO2 and, when indicated, treated with plate-bound anti-CD16 mAb (aCD16; 10 µg/ml; clone 3G8; produced in house) and/or IL-15 (10 ng/ml; Miltenyi). Anti-CD107a phycoerythrin was added at culture initiation. Brefeldin A (10 mg/ml; Sigma) was added for 4 hours. After 6 hours of stimulation, cells were stained for cell surface markers (CD3, CD8, CD45RA, and CCR7) and, when indicated, for intracellular cytokines (TNF-α and IFN-γ).

Antibody-Dependent Cellular Cytotoxicity Response

The antibody-dependent cellular cytotoxicity (ADCC) response of CD8 subsets was measured after challenge and culture with Raji cells coated with anti-CD20 antibody. CD8 T cells were cultured for 30 minutes with Raji cells precoated or not precoated with 10 mg/ml rituximab (Roche) at a ratio of 10:1 in 96-well, round-bottomed plates at 37°C in 5% CO2 in the presence of anti-CD107a phycoerythrin mAb. After 6 or 18 hours of stimulation, cells were stained for cell surface markers (CD3, CD8, CD45RA, and CCR7).

Binding of CD8 to Single-Antigen HLA Class II Beads

Purified CD8 T cells or cells from CD8 subsets (2×105–3×105) were incubated O/N with 2 µl of single-antigen HLA class II (OneLambda) beads previously incubated with 20 µl of serum from immunized KTx (treated with 0.1 mol/L EDTA for 10 min). When appropriate, cells were stained with anti-CD8, anti-CD45RA, and anti-CCR7 mAbs to identify CD8 subsets. A Boolean gating strategy was used to assess the frequency of cells in CD8 subsets bound to single-antigen HLA class II beads in the presence of the appropriate serum (Supplemental Figure 2).

RNA-Seq

Naive (CD45RA+CD28+), EM (CD45RACD28), and TEMRA (CD45RA+CD28) CD3+CD8+ T cells were FACS sorted (purity >95%) from freshly isolated human PBMCs obtained from eight healthy donors (obtained by the Etablissement Français du Sang). Cell pellets were resuspended in Buffer RLT (Qiagen) containing 1% β-mercaptoethanol before subsequent RNA extraction using an RNeasy Micro Kit according to the manufacturer’s instructions (Qiagen). The quality and quantity of the RNA were assessed by infrared spectrometry (Nanodrop) and an Agilent bioanalyzer (Agilent RNA 6000 Pico Kit). Smart-Seq2 libraries were prepared by the Broad Technology Labs and sequenced by the Broad Genomics Platform according to the SmartSeq2 protocol with some modifications.30 Briefly, total RNA was purified using RNA-SPRI beads, polyA+ mRNA was reverse transcribed to cDNA, and amplified cDNA was subject to transposon-based fragmentation that used dual indexing to barcode each fragment of each converted transcript with a combination of sample-specific barcodes. Sequencing was carried out as paired-end 2×25-bp reads with an additional eight cycles per index. Data were separated by barcode and aligned using TopHat version 2.0.10 with the default settings. Transcripts were quantified by the Broad Technology Labs computational pipeline using Cuffquant version 2.2.1.31 Briefly, data were processed through Cuffnorm if 50% of the reads aligned and if at least 100,000 pairs were aligned per sample. The default settings, including geometric normalization, were used for normalization and expression level data in the form of log2-transformed fragments per kilobase of transcript per million mapped fragment values were used for subsequent analyses. The selection of the most discriminative genes to classify the three CD8 subsets was performed using sparse partial least squares discriminant analysis (sPLS-DA)32 with the mixOmics package in R.33 A total of 100 repeats were used to establish the sPLS-DA model. Two components within each set of 100 genes were optimal to discriminate the three CD8 subsets. For gene expression representation, sample plots of the sPLS-DA and clustering analysis results were generated in R version 3.5.1 using the ade434,35 and heatmap3 (https://CRAN.R-project.org/package=heatmap3) packages, respectively. The biologic significance of selected genes was assessed using High-Throughput GoMiner.36 Gene-ontology categories enriched with a false discovery rate of <5% and containing at least five represented genes were selected. RNA-seq data were deposited in the Gene Expression Omnibus under the accession number GSE129356.

Statistical Analyses

Biomarkers were dichotomized with respect to the median value to avoid the log-linearity assumption (thus avoiding inflation of the type-1 error rate associated with the estimation of the optimal cutoff value for predicting graft survival). Graft survival was the primary outcome and was defined as the time between the transplantation and the return to dialysis or preemptive retransplantation (deaths were censored). Survival curves were obtained using the Kaplan–Meier estimator.37 The raw and KTFS-adjusted associations were calculated using Cox proportional hazards models.38 The Holm procedure was used to correct for the multiplicity of the tests. The CD8 phenotypes with a corrected P value <0.05 were selected for further analyses. The discriminatory capacities were evaluated by the AUC for data up to 8 or 11 years post-transplant obtained via the inverse probability censoring weighted estimator.39 The corresponding 95% confidence intervals (CIs) and P values related to the differences between AUC values were obtained by nonparametric bootstrap sampling (1000 iterations).

All statistical analyses were performed using R version 3.3.2 or GraphPad Prism. The package ROCt version 0.9 was used to generate the time-dependent ROC curves (www.labcom-risca.com/packages-r). The package nricens was used to calculate the net reclassification improvement. The package corrplot (https://github.com/taiyun/corrplot) was used to calculate and visualize the correlation between the CD8 cell–related populations. Mann–Whitney U tests, Kruskal–Wallis tests followed by Dunn post hoc tests, and paired Wilcoxon tests were used as appropriate, and the type of test used is included in the figure legends. Multiple comparisons were corrected using the two-stage linear step-up procedure of Benjamini et al.,40 and Q was set to 5%. All P values are given as exact values or as P<0.001.

Study Approval

PBMCs were prospectively collected from 284 KTx in the DIVAT biocollection (www.divat.fr) and stored in the Biologic Resource Center of the Nantes University Hospital, F-44093, France (BRIF: BB-0033-00040). All donors were informed of the final use of their blood and signed a written-informed-consent form. The University Hospital Ethical Committee and the Committee for the Protection of Patients from Biologic Risks approved the studies involving patients. Adults who received kidney grafts from deceased donors and had a functional transplant at the first anniversary of transplantation were included. Only patients without missing KTFS values were retained (i.e., patients for whom information about creatinemia at 3 and 12 months, proteinuria at 12 months, recipient sex and age, number of previous transplants, last donor creatinemia value, and number of rejection episodes during the first year post-transplant were available). Finally, the availability of frozen PBMCs at 12±6 months was used to select the patients.

Results

Characteristics of the Cohort

The demographic and clinical characteristics of the population are shown in Table 1. The KTx underwent transplantation between July 2003 and January 2012. Among the 284 KTx alive with a functional kidney graft at 1 year post-transplant (study baseline), 57 returned to dialysis at the end of the follow-up, and 39 died. The median follow-up time was 8.3 years. At 5 and 10 years, the graft survival rates were 91.5% (95% CI, 88.2% to 94.9%) and 75.3% (95% CI, 69.2% to 82.1%), respectively (Figure 1A). For 89.1% of the KTx, this kidney transplant was their first. The mean age of the recipients and donors was 51.1 years (range, 18–83 years) and 51.8 years (range, 6–82 years), respectively.

Table 1. - Description of the quantitative and qualitative characteristics at the study baseline (i.e., 12 mo post-transplant; n=284)
Characteristics Mean±SD or Number (%)
Recipient
 Age (yr) 51.07±13.21
 Body mass index (kg/m2) 24.28±4.12
 Cold ischemia time (h) 21.60±8.11
 Recipient creatinemia at 3 mo (μmol/L) 141.80±49.92
 Recipient creatinemia at 12 mo (μmol/L) 139.05±46.60
 Recipient proteinuria at 12 mo (g/24 h) 0.41±0.82
 KTFS 3.91±1.32
 First kidney transplant 253 (89.08)
 Sex (male) 180 (63.38)
 Recurrent initial disease 81 (28.52)
 CMV serology (positive) 148 (52.11)
 EBV serology (positive) 273 (96.81)
 HCV serology (positive) 10 (3.52)
 >4 HLA mismatches (A+B+DR) 49 (17.25)
 ≥1 Acute rejection episode during the first year 30 (10.56)
 Diabetes 46 (16.20)
 Arterial hypertension 268 (94.37)
 Cardiovascular history 107 (37.68)
 Cancer history 15 (5.28)
 Immunization with anti-HLA class I at day 0 53 (20.00)
 Immunization with anti-HLA class II at day 0 47 (17.87)
 Tacrolimus trough level (ng/ml) 8.62±3.54
Donor
 Age (yr) 51.76±16.36
 Last donor creatinemia (μmol/L) 87.61±63.38
 Male donor 169 (59.51)
 Vascular cause of donor death 163 (57.39)
CMV, cytomegalovirus; EBV, Epstein–Barr virus; HCV, hepatitis C virus.

fig1
Figure 1.:
A low frequency of EM CD8 T cells at 1 year post-transplant identifies patients with increased risk of graft failure. (A) Kidney graft survival in the overall cohort. Patients were included 12 months after transplantation, and the survival of their grafts was assessed using the Kaplan–Meier estimator. The number of patients at risk was calculated every year. (B) Graft survival curves according to the KTFS cutoff value. Patients were stratified according to the KTFS at 12 months post-transplant as low risk (KTFS≤4.17) or high risk (KTFS>4.17), and graft survival was assessed using the Kaplan–Meier estimator. The number of patients at risk was calculated every year. (C) Prognostic value of the KTFS. Time-dependent ROC curves up to 8 and 11 years post-transplant related to the KTFS (n=284). (D) Correlogram of CD8-related markers measured 12 months post-transplant. Correlations were tested using Spearman rank correlation, and only significant correlations are shown (P<0.01). (E) An EM CD8 T cell percentage of <36% was associated with increased risk of graft failure. Patients considered at high risk (KTFS>4.17) were stratified according to the median EM (CD45RACCR7) CD8 T cell percentage 12 months post-transplant, and graft survival was assessed using the Kaplan–Meier estimator. The number of patients at risk was calculated every year.

An Increase in the TEMRA/EM CD8 T Cell Ratio Identifies KTx at High Risk of Kidney-Graft Failure

We first validated the predictive value of the KTFS in this new cohort of 284 KTx. In terms of discriminatory capacity, the AUC at 8 years post-transplant was 0.75 (95% CI, 0.66 to 0.83) (Figure 1B), a prognostic value similar to that found in the original study (AUC, 0.78).21 Patients were stratified at 1 year post-transplant according to the previously defined KTFS value21 into the low-risk group (KTFS≤4.17) and the high-risk group (KTFS>4.17), and we demonstrated that the graft survival rate was significantly different between the two groups (HR, 1.26; P<0.001; Figure 1B). At 8 years post-transplant, the graft survival rates were 86.6% (95% CI, 80.5% to 93.1%) and 61.6% (95% CI, 49.8% to 76.3%) for patients with a low risk and a high risk of graft failure, respectively (Figure 1C).

We previously reported that an increase in highly differentiated TEMRA CD8 cells in patients with stable graft function for >5 years is a risk factor for graft dysfunction.19 Here, we assessed the association between the risk of kidney graft failure (a more stringent clinical criterion than that used in our previous study) and the early monitoring of CD8-related markers. Concurrent with the KTFS calculation, we monitored the frequency of CD8 T cell subsets using the phenotypic markers CD45RA and CCR7 (naive, CD45RA+CCR7+; TEMRA, CD45RA+CCR7; EM, CD45RACCR7; and central memory, CD45RACCR7+), markers of differentiation (CD27, CD28, and EOMES), the expression of cytotoxic molecules (GZMB and PERF-1), and markers associated with the secretion of proinflammatory cytokines (T-bet and CD57). After adjustment with the standard clinical metrics summarized in the KTFS, the 1-year frequency of TEMRA CD8 was associated with an increased risk of graft failure (HR, 1.61; P=0.083; Table 2). In contrast, an increased frequency of EM CD8 was associated with a reduced risk of graft failure (HR, 0.39; P=0.0012; Table 2). The increase frequency of EM CD8 was associated with a 10% higher frequency of anti-thymocyte globulin (ATG) as induction therapy (Supplemental Figure 3A). However, at the time of analysis, the maintenance therapy was similar across KTx (Supplemental Figure 3B). Finally, a higher frequency of chronic rejection in the KTx EMlow group as compared with the KTx EMhigh group was observed in for-cause kidney graft biopsy samples of the 57 patients with kidney graft failure (59% versus 42%, respectively; Supplemental Figure 3C). Collectively, our data suggest that the accumulation of TEMRA CD8 associated with a decrease in EM CD8 is not only associated with a higher risk of kidney graft failure but also that the accumulation of TEMRA CD8 is associated with a higher frequency of immune-related rejection.

Table 2. - Bivariate Cox models of the association of CD8-related biomarkers with return to dialysis
Marker Median HR P Value Corrected P Value
EM 36.02 0.39 0.001 0.027
TEMRA 31.70 1.61 0.083 >0.99
Naive 14.65 0.85 0.56 >0.99
CM 3.72 1.13 0.67 >0.99
GZMb+PERF-1 30.35 1.77 0.045 0.95
GZMbPERF-1 31.75 1.50 0.15 >0.99
GZMb+PERF-1+ 22.10 0.78 0.37 >0.99
CD28CD27+ 5.41 0.68 0.16 >0.99
CD28CD27 9.92 1.48 0.17 >0.99
CD28CD27 26.50 1.19 0.52 >0.99
CD28+CD27+ 42.90 0.94 0.84 >0.99
TBX21high 63.75 1.03 0.92 >0.99
EOMES+ 76.70 0.70 0.21 >0.99
TBX21+EOMES 7.47 0.76 0.33 >0.99
TBX21EOMES 19.50 1.25 0.43 >0.99
TBX21EOMES+ 20.55 1.22 0.48 >0.99
TBX21+EOMES+ 40.35 1.03 0.91 >0.99
TBX21CD57 34.35 0.89 0.67 >0.99
TBX21+CD57 21.15 0.89 0.68 >0.99
TBX21CD57+ 2.12 1.10 0.73 >0.99
TBX21+CD57+ 33.50 0.96 0.88 >0.99
HRs were adjusted for the KTFS. Corrected P values were obtained using the Holm method. CM, central memory.

The association between the TEMRA/EM CD8 proportion and kidney graft survival prompted us to hypothesize that the prognostic value of KTFS could be improved by combining the KTFS with the frequency of EM/TEMRA CD8 at 1 year post-transplant. As expected, a strong correlation was observed between the percentages of TEMRA and EM CD8 (P<0.001; Figure 1D), and we considered only the percentage of the EM CD8 (the lowest P value in Table 2). Among the patients at high risk of graft failure (KTFS>4.17; n=85), the risk of graft failure was 2.3-fold (95% CI, 1.1 to 4.9) higher in patients with an EM CD8 percentage <36% (median of EM CD8 in the cohort; n=41) than in patients with a higher EM CD8 percentage (n=44). Therefore, we demonstrated that, in patients at high risk of graft failure, a 1-year decreased percentage of EM CD8 T cells post-transplant is associated with the poorest prognosis. More precisely, in patients at high risk of graft failure (KTFS>4.17), the 7-year survival rate was 62.4% (95% CI, 58.2% to 87.6%) and 82.6% (95% CI, 75.5% to 97.0%) for patients with low or high EM CD8 percentages, respectively (Figure 1E). Collectively, our data demonstrate that the measurement of CD8 memory subsets at 1 year post-transplant improves the identification of patients who will later return to dialysis as a consequence of graft failure.

TEMRA and EM CD8 Exhibit Similar Effector Functions upon Donor-Specific Stimulation

Given the low TCRVβ repertoire diversity of TEMRA compared with that of EM CD8,18,20 we hypothesized that TEMRA CD8 are enriched in donor-specific reactive CD8 T cells, which could explain the inverse kidney graft outcomes between KTx stratified according to the TEMRA/EM CD8 ratio. Donor and recipient PBMCs were collected from 24 living-donor KTx before and 1-year after transplant. We first evaluated the consequence of kidney transplantation on the frequency and phenotype of CD8 subsets. The strong immune challenge induced by allogeneic kidney transplantation results in a decrease in naive CD8 T cells (31.67%±3.13% versus 23.60%±2.54% before and 1-year after transplant, respectively; Supplemental Figure 4A) and an increase in TEMRA CD8 (24.69%±3.76% versus 38.32%±4.06% before and 1-year after transplant, respectively; Supplemental Figure 4A). Native GZMB expression was restricted to TEMRA CD8 (Supplemental Figure 4B) and, as expected, expression of the TBX21 transcription factor and EOMES was restricted to the memory (EM and TEMRA) CD8 cell compartment (Supplemental Figure 4B). CD8 subsets were then purified from living-donor KTx and stimulated with donor-derived, T cell–depleted PBMCs. A strong upregulation of the early activation marker CD69 was observed in naive and memory (TEMRA and EM) CD8 after donor-specific stimulation (Figure 2A). However, the expression of the high-affinity IL-2Rα chain, CD25, and the cytotoxic marker CD107a was restricted to the memory CD8 subsets, and the magnitude of CD25 and CD107a expression did not differ between EM and TEMRA CD8 (Figure 2A). This early and memory-restricted activation profile was confirmed by analysis of culture supernatant from donor-specific CD8 subsets (Figure 2B). In addition, high levels of proinflammatory cytokines (IFN-γ, TNF-α, and IL-17A) and cytotoxic molecules (granulysin, PERF-1, GZMA, and sFASL) were found in the supernatant from both TEMRA and EM CD8 (Figure 2B). Finally, after donor-specific stimulation, vigorous proliferation of naive, EM, and TEMRA CD8 was observed (Figure 2C). Of note, syngeneic stimulation has previously been shown to fail to elicit the activation of CD8 T cells12 and, in agreement with a previous report,41 third-party stimulation elicits a strong upregulation of CD25, associated with the expression of CD107a, only in the memory CD8 compartment (Supplemental Figure 5). Collectively, these findings demonstrate that TEMRA and EM CD8 exhibit similar rapid and early functional responses to donor-specific stimulation.

fig2
Figure 2.:
TEMRA and EM CD8 T cells from KTx exhibit similar potent effector responses upon donor-specific stimulation. CD8 T subsets (naive, EM, and TEMRA) were purified from KTx before (M0) and 12 months after transplantation (M12) and stimulated with donor-specific, T cell–depleted PBMCs for (A and B) 48 hours or (C) 5 days. (A) Frequencies of CD8 subsets expressing activation (CD69 and CD25) or degranulation (CD107a) markers. Representative flow data and the gating strategy are shown (n=10). (B) Concentration of cytokines and cytotoxic molecules secreted by donor-specific stimulated CD8 subsets (n=10). (C) Proliferation of CD8 subsets was assessed according to the dilution of CPDeFluor450 signal within the subset of Annexin V cells. Representative flow data are shown (n=5–6). The bars indicate the mean±SD, and each point represents a single KTx. The P values were calculated using nonparametric ANOVA (Kruskal–Wallis) with the Dunn multiple comparisons test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

TEMRA CD8 Cells Exhibit an Innate-like Signature and Mediate ADCC upon CD16 Engagement

The similar responses of TEMRA and EM CD8 from KTx after donor-specific stimulation prompted us to identify alternative pathways of activation of TEMRA CD8 that could account for their pathogenic function leading to kidney graft failure. Because we previously showed that the immune function of TEMRA CD8 was similar between KTx and HV,20 eight sets of matched samples of naive, EM, and TEMRA CD8 isolated from the peripheral blood of HV were analyzed by RNA-seq to identify specific gene-expression profiles. Discriminatory gene analysis (sPLS-DA; see the Methods section) revealed that CD8 subsets could be efficiently identified and grouped (72.4% of the variance was explained by components 1 and 2; Figure 3A). The identification of CD8 subset-specific gene signatures was confirmed by unsupervised clustering of CD8-subset transcriptomes in the discriminatory gene analysis (Figure 3B). The top genes contributing to the identification of TEMRA CD8 were involved in cytotoxicity (GZMB, GNLY, and PERF-1), transport of lysosomal enzymes (GNPTAB), cell surface receptor signal transmission with key roles in the regulation of innate and adaptive immune responses (LYN), and—interestingly —innate immunity–related functions (KIR3DL1, KIR3DL2, KLRD1, CD244 or 2B4, CD300A, and FCGR3A or CD16) (Supplemental Figure 6, Supplemental Table 2). Higher expression of CD16 was observed at the transcriptome level in TEMRA CD8 than in naive and EM CD8 (Figure 3C); this finding was confirmed via analysis of the phenotype of CD3+CD8+CD16+ cells (61.1%±5.1% TEMRA versus 8.4%±2.6% EM; P<0.001; Figure 3D). Whereas most CD3+CD8+CD16+ cells exhibited a TEMRA phenotype, a fraction of TEMRA CD8 expressed CD16 (11.8%±3.9%; Figure 3E). To validate the expression of CD16 by TEMRA CD8 and to gain additional insight into these CD8 subsets, we analyzed data from CyTOF performed on PBMCs from HV.26Via unsupervised clustering (PhenoGraph; see the Methods section) and t-distributed stochastic neighbor embedding to visualize high-dimension data with single-cell resolution in two dimensions, we identified 23 clusters based on the expression of 24 markers (Supplemental Figure 7). Major CD8 subsets (naive, central memory, EM, and TEMRA) were identified according to the expected expression of CD45RA and CCR7 and were localized in neighboring locations on the t-distributed stochastic neighbor embedding map. Eight subsets of TEMRA CD8 were identified (clusters 5, 9, 10, 16, 17, 18, 22, and 23) with a shared phenotype (CD45RA+CCR7CD25CD62LCCR6GZMb+CD95+CD11a+CD57+). Most TEMRA CD8 subsets were CCR5+CCR6CCR4int. The TEMRA subsets differed in their expression of costimulatory molecules (CD28 and CD27), IL-7 receptor α-chain (CD127), and inhibitory molecules (BTLA, CTLA4, and LAG3). Interestingly, CD16 expression was restricted to TEMRA subsets, in agreement with the transcriptomic and flow cytometry data (Figure 3, A–E). Not only were TEMRA CD16+ (clusters 5, 9, 10, 16, 17, 18 and 22) and TEMRA CD16 (cluster 23) subsets identified, but the expression of the activated innate-like marker CD56 also differed among the TEMRA CD8 subsets (Supplemental Figure 7). In summary, unsupervised analysis of large-scale CyTOF data confirmed the existence of TEMRA CD8 subsets that could be characterized according to CD16 expression, suggesting a potential alternative pathway of TEMRA CD8 activation upon antibody-mediated Fc receptor engagement.

fig3
Figure 3.:
Selective activation of TEMRA CD8 T cells upon CD16 crosslinking. (A) PLS-DA sample plot and (B) heatmap showing scaled expression values of discriminating genes for CD8 T subsets (naive, EM, and TEMRA) purified from eight HV. (C) Expression of the CD16 transcript by CD8 T cell subsets. (D) Phenotype of CD3+CD8+CD16+ cells according to CD45RA and CCR7 expression (n=13). (E) Frequencies of CD16+ cells among CD3+CD8+ subsets. Representative flow data are shown (n=13). (F) Frequencies of CD8 subsets secreting TNF-α, IFN-γ, and CD107a after exposure to the indicated stimuli for 4 hours. Notably, naive CD8 T cells exhibited neither cytokine production nor degranulation. The bars indicate the mean±SEM of data from eight HV. (G) Expression of CD107a by CD8 subsets after 24 hours of coculture with Raji cells with or without rituximab (RTX). Representative flow data are shown (n=5). The bars indicate the mean±SEM, and each point represents a single HV. The P values were calculated using nonparametric ANOVA (Kruskal–Wallis) with (D and E) the Dunn multiple comparisons test or (F and G) a Wilcoxon matched-pairs signed-rank test. *P<0.05, **P<0.01, ***P<0.001. Dim, dimension.

To directly assess the functionality of CD16 in TEMRA CD8, purified naive, EM, and TEMRA CD8 were primed with plate-bound aCD16 (clone 3G8, mouse IgG1) for 4 hours, and the expression of the cytotoxic marker CD107a and production of IFN-γ and TNF-α were assessed (Figure 3F, Supplemental Figure 8, A–C). Short-term crosslinking of CD16 induces the activation of TEMRA CD8, as shown by increases in proinflammatory cytokine secretion and cytotoxicity compared with these parameters in EM CD8. Neither IFN-γ and TNF-α nor CD107a were found in naive CD8. Furthermore, a substantial increase in the effector response of TEMRA CD8 was observed in the presence of IL-15 (Figure 3F). Irrelevant coated antibodies (anti-CD4 mAb, mouse IgG1; anti-HLA class I mAb, mouse IgG2a) failed to elicit a cytotoxic response by TEMRA, demonstrating the specificity of the CD16-triggered activation (Supplemental Figure 8C). Finally, we measured the ability of TEMRA CD8 to mediate ADCC by challenging CD8 T cells with Raji cells coated with anti-CD20 mAb (Figure 3G). Degranulation was restricted to the TEMRA CD8 compartment (1.2%±0.4% versus 8.3%±3.2% CD107a+ for treatment with Raji cells versus treatment with Raji cells and anti-CD20 mAb, respectively; P=0.031). Similar results were obtained for shorter stimulation times (Supplemental Figure 8D). The degranulation of TEMRA CD8 induced by ADCC was CD16 dependent because blocking aCD16 was sufficient to prevent the cytotoxic response of TEMRA CD8, whereas irrelevant Ig (anti-HLA class I mAb) had no effect on the ADCC (Supplemental Figure 8E). These data demonstrate a unique innate-like signature in TEMRA CD8 and the ability to activate TEMRA CD8 by either the interaction between TCR and donor HLA/peptide complex or by the activation of CD16 upon Ig ligation.

CD16 Engagement Selectively Activates TEMRA CD8 from KTx

To investigate means by which chronic stimulation induced by allogeneic kidney transplantation modified the innate-like function of TEMRA CD8, IFN-γ and TNF-α production was examined after short-term in vitro crosslinking of CD16 on CD8 T cells purified from KTx. A significant and selective increase in the secretion of proinflammatory cytokines by TEMRA CD8 from KTx relative to this secretion by EM CD8 was observed (Figure 4A). The agonist effect of plate-bound aCD16 on TEMRA cells from KTx was further enhanced upon exposure to IL-15 (Figure 4A). The cytotoxic activity of TEMRA CD8 from KTx was similarly triggered upon combined exposure to aCD16 and IL-15 stimulation (Figure 4A). Only TEMRA CD8 from KTx mediated ADCC (0.93%±0.19% versus 4.15%±1.77% CD107a+ for treatment with Raji cells versus treatment with Raji cells and anti-CD20 mAb, respectively; P=0.016; Figure 4B). We ultimately hypothesized that TEMRA CD8 could selectively interact with HLA molecules, not only upon TCR-HLA interaction but also upon the binding of anti-HLA class II Ig to CD16 receptors expressed by TEMRA CD8. To test this hypothesis, we incubated CD8 T cells with single-antigen HLA class II beads in the presence of serum from either immunized KTx (with multiple HLA class II specificities) or from male HV without any known immunization (Figure 4C and Supplemental Figure 2). Exposure to immunized serum resulted in selective binding of TEMRA CD8 T cells to HLA class II molecules (percentage of CD8 bound to HLA class II molecules, 0.50%±0.13% versus 0.17%±0.04% for TEMRA and EM, respectively; P=0.004; Figure 4C). Finally, preincubation of TEMRA CD8 with blocking aCD16 prevents the subsequent interaction with DSA-coated HLA class II molecules (P=0.031; Figure 4D). Taken together, these results demonstrate that the activation of TEMRA CD8 from KTx could be achieved through either TCR or CD16 stimulation and could foster the inflammatory response and promote kidney graft failure.

fig4
Figure 4.:
Activation of the CD16 pathway selectively induces the proinflammatory response of TEMRA CD8 T cells from KTx. (A) Frequencies of CD8 subsets secreting TNF-α, IFN-γ, and CD107a after exposure to the indicated stimuli for 4 hours. Notably, naive CD8 T cells exhibited neither cytokine production nor degranulation. The bars indicate the mean±SEM of data from seven KTx. (B) Expression of CD107a by CD8 subsets after 24 hours of coculture with Raji cells with or without rituximab (RTX). Representative flow data are shown (n=6). The bars indicate the mean±SEM, and each point represents a single KTx. (C and D) Frequencies of CD3+CD8 subsets bound to (C) single-antigen HLA class II beads coated with serum from nonimmunized male individuals (serum control) or with serum from immunized KTx (n=9) or (D) before and after preincubation of TEMRA CD8 with aCD16 (n=6). The P values were calculated using nonparametric ANOVA (Kruskal–Wallis) with (A) the Dunn multiple comparisons test or (B) a Wilcoxon matched-pairs signed-rank test. *P<0.05, **P<0.01.

Discussion

The ability to stratify KTx according to the risk of graft failure is a major challenge. Identifying patients with a high risk of graft loss as early as possible may offer an early therapeutic window for intervention and, mainly, adaptation of the immunosuppressive drug regimen. Clinical-based scoring systems such as the KTFS21 form one approach to address this challenge. However, patients’ survival prospects will greatly benefit from biomarkers that combine the expectation of biomarker research (precision and sensitivity) with the cause of graft rejection. Here, we reported that monitoring EM/TEMRA CD8 in high-risk KTx (KTFS>4.17) enables the identification of KTx with an immunologic risk of kidney graft failure, because the graft survival rate was lower in KTx with a lower frequency of EM CD8 (i.e., a higher frequency of TEMRA CD8) than in KTx with a higher frequency of EM CD8. Consistent with our results obtained in KTx recruited >5 years after transplantation,19 these results showed that the modification of the CD8 compartment occurred at an earlier time point than originally thought and has a strong negative effect on kidney graft outcome. For the first time, we showed that KTx with a high risk of kidney graft failure can be identified based on the combination of a clinical metrics-based score with the monitoring of CD8 frequencies, a facile method in daily clinical practice.

A strong heterogeneity in the usage of the T cell compartment is observed in KTx with stable graft function, despite daily treatment with immunosuppressive drugs.16,18,19,42 The results of this study further confirmed this finding, showing high variability in the frequency of EM and TEMRA CD8 among KTx as early as 12 months post-transplant, a time point well into the induction therapy regimen. However, the increased frequency of EM CD8 was associated with a 10% higher frequency of ATG as induction therapy and further studies are needed to decipher the susceptibility of CD8 subsets to anti-thymocyte globulin and the relative contribution of ATG to the lower kidney graft failure. In addition, this heterogeneity demonstrates that, in some patients, the current immunosuppressive drugs failed to prevent the increase of TEMRA CD8. The lack of TEMRA CD8-specific therapeutics could be partially explained by the misunderstanding of TEMRA CD8 functionality. This CD8 subset has long been considered a hallmark of immune senescence, and the elderly and patients with chronic viral infections have historically been the major populations of interest.43,44 However, expansion of pathogenic TEMRA CD8 after allotransplantation has been documented in patients with autoimmune diseases (multiple sclerosis,45 lupus,46,47 ANCA,48 or primary Sjögren syndrome49).19,50 TEMRA CD8 exhibit a potent inflammatory response when appropriately stimulated. Here, we demonstrated that the response of TEMRA and EM CD8 does not differ after donor-specific stimulation; both memory CD8 populations favor the generation of a rapid inflammatory response characterized by cytotoxic function (CD107a upregulation and high levels of the cytotoxic molecules GZMA, GNLY, and PERF-1) and the secretion of a wide spectrum of proinflammatory cytokines (IFN-γ, TNF-α, and IL-17A) upon donor-specific stimulation. One hypothesis explaining the negative effect of TEMRA CD8 on long-term graft survival could be related to an accumulation of preformed cytotoxic molecules and proinflammatory cytokines (RNA-seq data and19) within TEMRA CD8 that could be released upon TCR stimulation via the direct pathway of allorecognition, which is maintained for years after transplantation and is negatively correlated with graft function.51 In addition to identifying the potent TCR-dependent response of TEMRA CD8, our report clarifies a unique feature restricted to TEMRA CD8 that could account for the poor clinical outcomes in KTx with a high frequency of TEMRA CD8. TEMRA CD8 express a transcriptomic signature associated with the innate-like population, including the expression of CD16 (FcγRIIIa). The engagement of CD16 results in the selective activation of TEMRA CD8, characterized by proinflammatory cytokine secretion and cytotoxic responses. Self-specific activated CD44hi CD8 mouse T cells were reported to express functional CD16 upon activation with IL-2 or with Ag and Il-2,52 as shown by the ability of self-specific CD8 to mediate ADCC independent of the TCR.52 The level of CD16 expression by NK cells is negatively regulated by CD3ζ,53 a subunit component of the FcγR CD16 and TCR/CD3 complex.54 For example, induced expression of CD3ζ in murine NK cells enhances the formation of CD3ζ/FcRγ heterodimers, prevents association with CD16,52,53 and decreases their ADCC function, suggesting that a delicate balance of CD16/CD3ζ is required to limit the expression of CD16. Our RNA-seq results revealed that TEMRA CD8 exhibited upregulated expression of the genes encoding the signaling adaptor FceRIγ, an ITAM-bearing adaptor molecule known to regulate cell surface CD16 expression and function, and its downstream signaling molecules 1-phosphatidylinositol-4,5-bisphosphate phosphodiesterase γ-2 (PLCG2) and zinc finger and BTB domain–containing protein 16 encoding the promyelocytic leukemia zinc finger (ZBTB16), which interact with the FceRIγ promoter55). We also found decreased expression of the transcription factor Bcl11b, which is reported to protect T-cell identity.56 The regulation of CD16 expression on TEMRA CD8, the identification of signaling events leading to CD16 expression, and the interconnection between the TCR signaling complex and CD16/FcγR require further investigation. However, our data support the concept of CD8 T-cell plasticity with the acquisition of innate-like functions by TEMRA CD8. With a dual activation mechanism that relies on TCR and FcγR signals, TEMRA CD8 are likely to promote and sustain an inflammatory environment leading to kidney graft failure.

Few reports have highlighted the expression of FcγR by human T cells. FcγRIIIa (CD16) expression was reported on human CD4 T cells from HV and patients with SLE, and its binding to immune complexes induces high secretion of IFN-γ.57 Memory CD8 T cells generated by viral or bacterial infection were shown to selectively express FcγRIIB and, upon engagement, contribute to inhibiting the cytotoxicity of memory CD8 T cells and their expansion after homologous challenge.58 In addition, interaction between T cells and FcγR-expressing antigen-presenting cells was also shown to result in the acquisition of FcγR by T cells via trogocytosis.59 Despite the scarcity of reports of FcγR expression by human T cells, the expression of NK-related markers such as KIRs and NKG2A by CD8αβ cells has long been reported.60,61 Increasing evidence is available regarding both the expression of innate-associated markers by naive62 and memory63–65 CD8 T cells and the identification of innate-like CD8 T cells. KIR/NKG2A+ CD8 T cells—identified in the blood of HV,64 in cord blood,64 and in patients with chronic myeloid leukemia65—rapidly produce IFN-γ in response to IL-12 and IL-18 stimulation and exhibit antigen-independent cytotoxic function. Exposure to IL-15 not only maintains the expression of NKp30 in NKp30+CD8 T cells but also promotes the differentiation of NKp30+CD8 T cells with concomitant acquisition of other NK receptors, high expression of T-bet, and low expression of the transcription factor Bcl11b and exhaustion markers (PD-1, CTLA-4, and Lag-3).62 We and others have shown that IL-15 regulates the function and homeostasis of TEMRA CD8 T cells.20,66,67 Signaling via IL-2 and IL-15 induces the loss of CD28,66,68 a phenotypic characteristic of TEMRA CD842,69; moreover, the shared IL-2 and IL-15 receptor β-chain CD122 was shown to be critical for costimulation-independent T cell alloreactivity.70 Selective inhibition of IL-15 signaling in TEMRA CD8 could thus be an appealing therapeutic strategy to limit the expansion and activation of TEMRA CD8 and, therefore, could lead to an improvement in long-term graft outcomes.

In summary, we hypothesize that TEMRA CD8 migrate to the graft, where donor HLA promotes their activation upon recognition by TCRs or engagement of the CD16 pathway by anti-HLA antibodies, thus favoring the induction of a sustained inflammatory microenvironment, including endothelial activation.20 This original mechanism of activation suggests that TEMRA CD8 are involved in cellular and humoral rejection of kidney grafts and that KTx will benefit from the monitoring of CD8 T cell subsets and the development of therapeutics specifically targeting TEMRA CD8.

Disclosures

None.

Funding

Dr. Yap is supported by a Fondation ProGreffe international fellowship grant. Dr. Jacquemont received a Société Française de Transplantation grant. This work was funded by ITMO Santé Publique grant A13053NS, supported by Agence Nationale de la Recherche (ANR; French National Research Agency) grant ANR-11-JSV1-0008-01, and supported in part by Réseau Thématique de Recherche et de Soins (RTRS) Fondation de Coopération Scientifique CENTAURE grants PAC8 and PAC9. This work was performed as a part of the IHU-CESTI project, which received French government financial support managed by the ANRvia the “Investment into the Future” program ANR-10-IBHU-005. The IHU-CESTI project is also supported by the Nantes Metropolis and the Conseil Régional des Pays de la Loire. This work was also supported by the FP7 VISICORT project, which has received funding from the European Union’s Seventh Framework Programme for research, technological development, and demonstration under grant agreement 602470. This work was performed as a part of the LabEX IGO program supported by the ANRvia the Investment into the Future program ANR-11-LABX-0016-01. This work was supported in the context of the ANR project BIKET (ANR-17-CE17-0008).

Published online ahead of print. Publication date available at www.jasn.org.

We would like to thank Dr. Bertram Bengsch and Dr. E. John Wherry for kindly sharing the CyTOF data and Samuel Granjeaud for his invaluable help with the R-based cytometric analysis.

The Biological Resource Center of the Nantes University Hospital, F-44093, France (BRIF: BB-0033-00040) guarantees the quality of the biologic samples.

Dr. Jacquemont and Ms. Tilly designed the experiments, performed the experiments, and analyzed the data. Dr. Yap and Dr. Doan-Ngoc performed the experiments and analyzed the data. Dr. Danger analyzed the transcriptomic data. Mr. Guerif and Dr. Giral assisted with human sample collection and processing, with patient consent, and they obtained ethical approval for human studies. Dr. Delbos and Mr. Martinet provided critical reagents. Dr. Foucher contributed to biostatistical data analyses. Dr. Degauque, Dr. Brouard, Dr. Giral, Dr. Foucher, and Dr. Jacquemont designed and supervised the study. Dr. Degauque performed the experiments, analyzed the data, and wrote the manuscript with input from all authors.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2019080847/-/DCSupplemental.

Supplemental Table 1. Gene ontology (GO) categories of CD8 subset gene signatures identified by sparse partial least squares discriminant analysis (sPLS-DA).

Supplemental Table 2. List of monoclonal antibodies used.

Supplemental Figure 1. The NK gene signature in kidney biopsies from KTx with ABMR is not restricted to the NK compartment.

Supplemental Figure 2. Gating strategy to analyze the binding of CD8 subsets with serum-coated single-antigen HLA class II beads.

Supplemental Figure 3. Kidney biopsies from patients with graft failure and low frequency of EM CD8 display higher rate of ABMR and TCMR.

Supplemental Figure 4. TEMRA CD8 T cells from living donor KTx and deceased donor KTx exhibit similar phenotypes.

Supplemental Figure 5. TEMRA and EM CD8 T cells from KTx exhibit similar potent effector responses upon third party stimulation.

Supplemental Figure 6. Unique gene signatures discriminate CD8 subsets via PLS-DA.

Supplemental Figure 7. Phenotype of CD16-expressing CD8 T cells using high-dimensional data analysis.

Supplemental Figure 8. Early and selective activation of TEMRA CD8 T cells upon CD16 crosslinking.

References

1. Loupy A, Lefaucheur C, Vernerey D, Prugger C, Duong van Huyen JP, Mooney N, et al.: Complement-binding anti-HLA antibodies and kidney-allograft survival. N Engl J Med 369: 1215–1226, 201324066742
2. Lefaucheur C, Loupy A, Vernerey D, Duong-Van-Huyen J-P, Suberbielle C, Anglicheau D, et al.: Antibody-mediated vascular rejection of kidney allografts: A population-based study. Lancet 381: 313–319, 201323182298
3. Roufosse C, Simmonds N, Clahsen-van Groningen M, Haas M, Henriksen KJ, Horsfield C, et al.: A 2018 reference guide to the banff classification of renal allograft pathology. Transplantation 102: 1795–1814, 201830028786
4. Aubert O, Loupy A, Hidalgo L, Duong van Huyen J-P, Higgins S, Viglietti D, et al.: Antibody-mediated rejection due to preexisting versus de novo donor-specific antibodies in kidney allograft recipients. J Am Soc Nephrol 28: 1912–1923, 201728255002
5. Lefaucheur C, Viglietti D, Bentlejewski C, Duong van Huyen J-P, Vernerey D, Aubert O, et al.: IgG donor-specific anti-human HLA antibody subclasses and kidney allograft antibody-mediated injury. J Am Soc Nephrol 27: 293–304, 201626293822
6. Bouquegneau A, Loheac C, Aubert O, Bouatou Y, Viglietti D, Empana J-P, et al.: Complement-activating donor-specific anti-HLA antibodies and solid organ transplant survival: A systematic review and meta-analysis. PLoS Med 15: e1002572, 201829799874
7. Hidalgo LG, Sis B, Sellarés J, Campbell PM, Mengel M, Einecke G, et al.: NK cell transcripts and NK cells in kidney biopsies from patients with donor-specific antibodies: Evidence for NK cell involvement in antibody-mediated rejection. Am J Transplant 10: 1812–1822, 201020659089
8. Hidalgo LG, Sellarés J, Sis B, Mengel M, Chang J, Halloran PF: Interpreting NK cell transcripts versus T cell transcripts in renal transplant biopsies. Am J Transplant 12: 1180–1191, 201222390872
9. Loupy A, Lefaucheur C, Vernerey D, Chang J, Hidalgo LG, Beuscart T, et al.: Molecular microscope strategy to improve risk stratification in early antibody-mediated kidney allograft rejection. J Am Soc Nephrol 25: 2267–2277, 201424700874
10. Lefaucheur C, Viglietti D, Hidalgo LG, Ratner LE, Bagnasco SM, Batal I, et al.: Complement-activating Anti-HLA antibodies in kidney transplantation: Allograft gene expression profiling and response to treatment. J Am Soc Nephrol 29: 620–635, 201829042454
11. Poggio ED, Clemente M, Riley J, Roddy M, Greenspan NS, Dejelo C, et al.: Alloreactivity in renal transplant recipients with and without chronic allograft nephropathy. J Am Soc Nephrol 15: 1952–1960, 200415213286
12. van de Berg PJ, Yong SL, Koch SD, Lardy N, van Donselaar-van der Pant KA, Florquin S, et al.: Characteristics of alloreactive T cells measured before renal transplantation. Clin Exp Immunol 168: 241–250, 201222471286
13. Ashton-Chess J, Dugast E, Colvin RB, Giral M, Foucher Y, Moreau A, et al.: Regulatory, effector, and cytotoxic T cell profiles in long-term kidney transplant patients. J Am Soc Nephrol 20: 1113–1122, 200919357258
14. Homs S, Mansour H, Desvaux D, Diet C, Hazan M, Buchler M, et al.: Predominant Th1 and cytotoxic phenotype in biopsies from renal transplant recipients with transplant glomerulopathy. Am J Transplant 9: 1230–1236, 200919422348
15. Obata F, Yoshida K, Ohkubo M, Ikeda Y, Taoka Y, Takeuchi Y, et al.: Contribution of CD4+ and CD8+ T cells and interferon-gamma to the progress of chronic rejection of kidney allografts: The Th1 response mediates both acute and chronic rejection. Transpl Immunol 14: 21–25, 200515814278
16. Baeten D, Louis S, Braud C, Braudeau C, Ballet C, Moizant F, et al.: Phenotypically and functionally distinct CD8+ lymphocyte populations in long-term drug-free tolerance and chronic rejection in human kidney graft recipients. J Am Soc Nephrol 17: 294–304, 200616338967
17. Brouard S, Le Bars A, Dufay A, Gosselin M, Foucher Y, Guillet M, et al.: Identification of a gene expression profile associated with operational tolerance among a selected group of stable kidney transplant patients. Transpl Int 24: 536–547, 201121457359
18. Miqueu P, Degauque N, Guillet M, Giral M, Ruiz C, Pallier A, et al.: Analysis of the peripheral T-cell repertoire in kidney transplant patients. Eur J Immunol 40: 3280–3290, 201021061447
19. Yap M, Boeffard F, Clave E, Pallier A, Danger R, Giral M, et al.: Expansion of highly differentiated cytotoxic terminally differentiated effector memory CD8+ T cells in a subset of clinically stable kidney transplant recipients: A potential marker for late graft dysfunction. J Am Soc Nephrol 25: 1856–1868, 201424652799
20. Tilly G, Doan-Ngoc T-M, Yap M, Caristan A, Jacquemont L, Danger R, et al.: IL-15 harnesses pro-inflammatory function of TEMRA CD8 in kidney-transplant recipients. Front Immunol 8: 778, 201728713392
21. Foucher Y, Daguin P, Akl A, Kessler M, Ladrière M, Legendre C, et al.: A clinical scoring system highly predictive of long-term kidney graft survival. Kidney Int 78: 1288–1294, 201020861817
22. Kaboré R, Haller MC, Harambat J, Heinze G, Leffondré K: Risk prediction models for graft failure in kidney transplantation: A systematic review. Nephrol Dial Transplant 32[Suppl 2]: ii68–ii76, 201728206633
23. Foucher Y, Meurette A, Daguin P, Bonnaud-Antignac A, Hardouin J-B, Chailan S, et al.: A personalized follow-up of kidney transplant recipients using video conferencing based on a 1-year scoring system predictive of long term graft failure (TELEGRAFT study): Protocol for a randomized controlled trial. BMC Nephrol 16: 6, 201525631635
24. Braun WE, Schold JD: Transplantation: Strength in numbers-predicting long-term transplant outcomes. Nat Rev Nephrol 7: 135–136, 201121283137
25. Moore J, He X, Liu X, Shabir S, Ball S, Cockwell P, et al.: Mortality prediction after kidney transplantation: Comparative clinical use of 7 comorbidity indices. Exp Clin Transplant 9: 32–41, 201121605021
26. Bengsch B, Ohtani T, Khan O, Setty M, Manne S, O’Brien S, et al.: Epigenomic-guided mass cytometry profiling reveals disease-specific features of exhausted CD8 T cells. Immunity 48: 1029–1045.e5, 201829768164
27. Becher B, Schlitzer A, Chen J, Mair F, Sumatoh HR, Teng KW, et al.: High-dimensional analysis of the murine myeloid cell system. Nat Immunol 15: 1181–1189, 201425306126
28. Chen H, Lau MC, Wong MT, Newell EW, Poidinger M, Chen J: Cytofkit: A bioconductor package for an integrated mass cytometry data analysis pipeline. PLOS Comput Biol 12: e1005112, 201627662185
29. Wong MT, Chen J, Narayanan S, Lin W, Anicete R, Kiaang HT, et al.: Mapping the diversity of follicular helper T cells in human blood and tonsils using high-dimensional mass cytometry analysis. Cell Reports 11: 1822–1833, 201526074076
30. Trombetta JJ, Gennert D, Lu D, Satija R, Shalek AK, Regev A: Preparation of single-cell RNA-Seq libraries for next generation sequencing. Curr Protoc Mol Biol 107: 1–17, 201424984854
31. Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, et al.: Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc 7: 562–578, 201222383036
32. Lê Cao KA, Boitard S, Besse P: Sparse PLS discriminant analysis: Biologically relevant feature selection and graphical displays for multiclass problems. BMC Bioinformatics 12: 253, 201121693065
33. Rohart F, Gautier B, Singh A, Lê Cao KA: mixOmics: An R package for ‘omics feature selection and multiple data integration. PLOS Comput Biol 13: e1005752, 201729099853
34. Bougeard S, Dray S: Supervised multiblock analysis in Rwith the ade4Package. J Stat Softw 86: 1–17, 2018
35. Dray S, Dufour A-B: The ade4 Package: Implementing the duality diagram for ecologists. J Stat Softw 22: 1–20, 2007
36. Zeeberg BR, Qin H, Narasimhan S, Sunshine M, Cao H, Kane DW, et al.: High-Throughput GoMiner, an ‘industrial-strength’ integrative gene ontology tool for interpretation of multiple-microarray experiments, with application to studies of Common Variable Immune Deficiency (CVID). BMC Bioinformatics 6: 168, 200515998470
37. Kaplan E, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: 457–481, 1958
38. Cox D: Regression models and life-tables. J Roy Stat Soc Series B Stat Methodol 34: 187–220, 1972
39. Blanche P, Dartigues J-F, Jacqmin-Gadda H: Review and comparison of ROC curve estimators for a time-dependent outcome with marker-dependent censoring. Biom J 55: 687–704, 201323794418
40. Benjamini Y, Krieger AM, Yekutieli D: Adaptive linear step-up procedures that control the false discovery rate. Biometrika 93: 491–507, 2006
41. Fischer M, Leyking S, Schäfer M, Elsäßer J, Janssen M, Mihm J, et al.: Donor-specific alloreactive T cells can be quantified from whole blood, and may predict cellular rejection after renal transplantation. Eur J Immunol 47: 1220–1231, 201728426152
42. Yap M, Tilly G, Giral M, Brouard S, Degauque N: Benefits of using CD45RA and CD28 to investigate CD8 subsets in kidney transplant recipients. Am J Transplant 16: 999–1006, 201626820487
43. Wertheimer AM, Bennett MS, Park B, Uhrlaub JL, Martinez C, Pulko V, et al.: Aging and cytomegalovirus infection differentially and jointly affect distinct circulating T cell subsets in humans. J Immunol 192: 2143–2155, 201424501199
44. Khan N, Shariff N, Cobbold M, Bruton R, Ainsworth JA, Sinclair AJ, et al.: Cytomegalovirus seropositivity drives the CD8 T cell repertoire toward greater clonality in healthy elderly individuals. J Immunol 169: 1984–1992, 200212165524
45. Salou M, Garcia A, Michel L, Gainche-Salmon A, Loussouarn D, Nicol B, et al.: Expanded CD8 T-cell sharing between periphery and CNS in multiple sclerosis. Ann Clin Transl Neurol 2: 609–622, 201526125037
46. McKinney EF, Lyons PA, Carr EJ, Hollis JL, Jayne DR, Willcocks LC, et al.: A CD8+ T cell transcription signature predicts prognosis in autoimmune disease. Nat Med 16: 586–591, 201020400961
47. McKinney EF, Lee JC, Jayne DR, Lyons PA, Smith KG: T-cell exhaustion, co-stimulation and clinical outcome in autoimmunity and infection. Nature 523: 612–616, 201526123020
48. Néel A, Bucchia M, Néel M, Tilly G, Caristan A, Yap M, et al.: Dampening of CD8+ T cell response by B cell depletion therapy in antineutrophil cytoplasmic antibody–associated vasculitis. Arthritis Rheumatol 71[4]: 641–650, 2019 30375745
49. Tasaki S, Suzuki K, Nishikawa A, Kassai Y, Takiguchi M, Kurisu R, et al.: Multiomic disease signatures converge to cytotoxic CD8 T cells in primary Sjögren’s syndrome. Ann Rheum Dis 76: 1458–1466, 201728522454
50. Reinke S, Geissler S, Taylor WR, Schmidt-Bleek K, Juelke K, Schwachmeyer V, et al.: Terminally differentiated CD8⁺ T cells negatively affect bone regeneration in humans [published correction appears in Sci Transl Med 5: 187er4, 2013]. Sci Transl Med 5: 177ra36, 201323515078
51. Bestard O, Nickel P, Cruzado JM, Schoenemann C, Boenisch O, Sefrin A, et al.: Circulating alloreactive T cells correlate with graft function in longstanding renal transplant recipients. J Am Soc Nephrol 19: 1419–1429, 200818417724
52. Dhanji S, Tse K, Teh H-S: The low affinity Fc receptor for IgG functions as an effective cytolytic receptor for self-specific CD8 T cells. J Immunol 174: 1253–1258, 200515661880
53. Arase H, Suenaga T, Arase N, Kimura Y, Ito K, Shiina R, et al.: Negative regulation of expression and function of Fc gamma RIII by CD3 zeta in murine NK cells. J Immunol 166: 21–25, 200111123272
54. Kurosaki T, Gander I, Ravetch JV: A subunit common to an IgG Fc receptor and the T-cell receptor mediates assembly through different interactions. Proc Natl Acad Sci U S A 88: 3837–3841, 19911827205
55. Schlums H, Cichocki F, Tesi B, Theorell J, Béziat V, Holmes TD, et al.: Cytomegalovirus infection drives adaptive epigenetic diversification of NK cells with altered signaling and effector function. Immunity 42: 443–456, 201525786176
56. Li P, Burke S, Wang J, Chen X, Ortiz M, Lee S-C, et al.: Reprogramming of T cells to natural killer-like cells upon Bcl11b deletion. Science 329: 85–89, 201020538915
57. Chauhan AK, Chen C, Moore TL, DiPaolo RJ: Induced expression of FcγRIIIa (CD16a) on CD4+ T cells triggers generation of IFN-γhigh subset. J Biol Chem 290: 5127–5140, 201525556651
58. Starbeck-Miller GR, Badovinac VP, Barber DL, Harty JT: Cutting edge: Expression of FcγRIIB tempers memory CD8 T cell function in vivo. J Immunol 192: 35–39, 201424285839
59. Hudrisier D, Clemenceau B, Balor S, Daubeuf S, Magdeleine E, Daëron M, et al.: Ligand binding but undetected functional response of FcR after their capture by T cells via trogocytosis. J Immunol 183: 6102–6113, 200919841164
60. Mingari MC, Moretta A, Moretta L: Regulation of KIR expression in human T cells: A safety mechanism that may impair protective T-cell responses. Immunol Today 19: 153–157, 19989577090
61. Mingari MC, Vitale C, Cambiaggi A, Schiavetti F, Melioli G, Ferrini S, et al.: Cytolytic T lymphocytes displaying natural killer (NK)-like activity: Expression of NK-related functional receptors for HLA class I molecules (p58 and CD94) and inhibitory effect on the TCR-mediated target cell lysis or lymphokine production. Int Immunol 7: 697–703, 19957547697
62. Correia MP, Stojanovic A, Bauer K, Juraeva D, Tykocinski L-O, Lorenz H-M, et al.: Distinct human circulating NKp30+FcεRIγ+CD8+ T cell population exhibiting high natural killer-like antitumor potential. Proc Natl Acad Sci U S A 115: E5980–E5989, 201829895693
63. Kim J, Chang D-Y, Lee HW, Lee H, Kim JH, Sung PS, et al.: Innate-like cytotoxic function of bystander-activated CD8+ T cells is associated with liver injury in acute hepatitis A. Immunity 48: 161–173.e5, 201829305140
64. Jacomet F, Cayssials E, Basbous S, Levescot A, Piccirilli N, Desmier D, et al.: Evidence for eomesodermin-expressing innate-like CD8(+) KIR/NKG2A(+) T cells in human adults and cord blood samples. Eur J Immunol 45: 1926–1933, 201525903796
65. Jacomet F, Cayssials E, Barbarin A, Desmier D, Basbous S, Lefèvre L, et al.: The Hypothesis of the human iNKT/Innate CD8(+) T-cell axis applied to cancer: Evidence for a deficiency in chronic myeloid leukemia. Front Immunol 7: 688, 201728138330
66. Chiu WK, Fann M, Weng NP: Generation and growth of CD28nullCD8+ memory T cells mediated by IL-15 and its induced cytokines. J Immunol 177: 7802–7810, 200617114451
67. Setoguchi R: IL-15 boosts the function and migration of human terminally differentiated CD8+ T cells by inducing a unique gene signature. Int Immunol 28: 293–305, 201626857736
68. Miyagawa F, Tagaya Y, Kim BS, Patel HJ, Ishida K, Ohteki T, et al.: IL-15 serves as a costimulator in determining the activity of autoreactive CD8 T cells in an experimental mouse model of graft-versus-host-like disease. J Immunol 181: 1109–1119, 200818606663
69. Appay V, Dunbar PR, Callan M, Klenerman P, Gillespie GM, Papagno L, et al.: Memory CD8+ T cells vary in differentiation phenotype in different persistent virus infections. Nat Med 8: 379–385, 200211927944
70. Mathews DV, Dong Y, Higginbotham LB, Kim SC, Breeden CP, Stobert EA, et al.: CD122 signaling in CD8+ memory T cells drives costimulation-independent rejection. J Clin Invest 128: 4557–4572, 201830222140
Keywords:

lymphocytes; chronic allograft rejection; clinical immunology; immunology and pathology; kidney transplantation

Copyright © 2020 by the American Society of Nephrology