Patients prior to RT in the period from August 2015 to March 2016 were suitable for participation. Each patient gave written informed consent to take part in this study. The study was approved by the Medical Ethical Committee of the Erasmus MC (MEC number 2015-301) and was conducted in accordance with the Declaration of Helsinki and the Declaration of Istanbul.

Clinical variables were assessed as shown in Table 1, including age of the recipient and donor, gender, CMV-seropositivity, human leukocyte antigen (HLA) class I and class II mismatches, current and historical panel reactive antibody (PRA) scores, number of RTs, warm ischemia time, number of living donor transplants, cause of chronic kidney disease, preemptive RT (defined as receiving a kidney before initiation of renal replacement therapy), number of related RT (receiving a kidney from a genetically related donor), and type of rejection. The HLA-typing was assessed according to international standards (American Society for Histocompatibility and Immunogenetics/the European Federation for Immunogenetics) using serologic and DNA-based techniques. The PRAs were determined at the laboratory of the blood bank in Leiden, the Netherlands.

Early acute rejection was defined as development of biopsy-proven acute allograft rejection according to the Banff criteria (14, 15) within 3 months after RT.

Peripheral blood mononuclear cells (PBMCs) and lymph node mononuclear cells (LNMCs) were isolated as described previously (12). Briefly, PBMCs were isolated from heparinized blood samples by using Ficoll-Paque Plus (GE healthcare, Uppsala, Sweden). Blood was drawn from renal transplant recipients 1 day before RT.

The waste material from renal transplant recipients removed during the implantation of the renal allograft was used to collect iliac LNs. The LNs were cut into small pieces and transferred onto a 70-µm sieve. A cell suspension was obtained after mashing the small pieces through this sieve. The cell suspension was then washed multiple times to eliminate remaining fat tissue. The isolated PBMCs and LNMCs were stored at −150°C with a minimum amount of 10 × 10⁶ cells/vial. Median yield of the LN amounted to 20 × 10⁶ cells.

Whole blood and freshly isolated LNMCs were used to assess T-cell differentiation status by flowcytometry. Differentiation status was based upon the study by Sallusto et al. (16), as described in detail previously (17). Briefly, expression of CD45RO and CCR7 was used to determine naïve and memory T-cell populations. Memory T-cell subsets were further defined as central memory (CM) T cells (CD45RO⁺CCR7⁺ T cells), effector memory (EM) T cells (CD45RO⁺CCR7⁻ T cells), and EMRA T cells (CD45RA⁺CCR7⁻ EM T cells) (16). In addition, CD28 expression was analyzed as this is lost upon T-cell differentiation (18–21). Cells were stained as described previously (2), and were measured on the FACSCanto II (BD) and analyzed with the fluorescence-activated cell sorting (FACS) Diva software version 6.1.2 (BD).

To dissect T cells into early- and late-differentiated cells, expression of CD28, CD27, CD57, and PD-1 was also analyzed. Naïve and CM T cells express CD28 and CD27. EM and EMRA T cells progressively lose expression of these molecules [CD4⁺ T cells first lose CD27 and then CD28, while the opposite is true for CD8⁺ T cells (22–24)] and express CD57 and PD-1 (25–28). Thus, early-differentiated T cells were defined as CD28⁺CD27⁺, CD28⁺PD-1⁻, and CD28⁺CD57⁻ T cells, while CD28^nullCD27⁻, CD28^nullPD-1⁺, and CD28^nullCD57⁺ T cells were defined as late-differentiated T cells. This measurement was performed on the Navios flow cytometer (Beckman Colter) and collected data were evaluated with the Kaluza software version 1.3 (Beckman Colter). T cells were analyzed with DuraClone IM T-cell subsets tubes (Beckman Colter) as described previously (12).

Relative telomere length of CD4⁺ and CD8⁺ T cells was determined by using flow fluorescent in situ hybridization on thawed PBMCs and LNMCs, as described in detail previously (17).

Recent thymic emigrants (RTEs) were defined as naïve T cells expressing CD31 and were assessed by flow cytometry, as described previously (29). T-cell receptor excision circle (TREC) content was determined using 1 × 10⁶ snap-frozen PBMCs and LNMCs. DNA was isolated from these snap-frozen samples and TREC content was detected using quantitative polymerase chain reaction as described previously (30). The TREC content is depicted as 1/ΔCT.

Peripheral blood mononuclear cells and lymph node mononuclear cells from renal transplant recipients (responders) were thawed and rested overnight. Then PBMCs and LNMCs were labeled with carboxyfluorescein succinimidyl ester (CFSE) (Molecular Probes^®, Leiden, the Netherlands) according to manufacturer’s instructions and stimulated in triplicate at 5 × 10⁴/well with irradiated PBMCs (40 Gy) of their corresponding donor, at a 1:1 ratio for 6 days. As a negative control, responders were stimulated with their own irradiated PBMCs or LNMCs (autologous stimulation). Responder cells were stimulated with phytohemagglutinin (PHA) 5 µg/ml to examine their maximum proliferative potential. On day 6, wells of the same condition were pooled and supernatant stored at −80°C. Proliferation was analyzed by measuring CFSE dilution and determining the frequency of CFSE⁻ cells. For this purpose, cells were stained using the following antibodies: AmCyan-labeled anti-CD3 (BD), pacific blue (PacB)-labeled anti-CD4 (BD), APC-Cy7-labeled anti-CD8 (BD), phycoerythrin (PE)-Cy7-labeled anti-CCR7 (BD Pharmigen), APC-labeled anti-CD45RO (BD), and PE-labeled anti-CD28 (BD). A dump-channel was applied to exclude unwanted cells from the analysis, by co-staining cells for the live-dead marker 7-AAD, peridin chlorophyll protein (PerCP)-labeled anti-CD19 (BD), PerCP-Cy5.5-labeled anti-CD56 (Biolegend), and PerCP-labeled anti-CD14 (BD) (Figure S1 in Supplementary Material). Samples were measured on the FACSCanto II (BD) and analyzed using FACS Diva software version 6.1.2 (BD).

Concentrations of IFN-γ, tumor necrosis factor-alpha (TNF-α), and granzyme B were determined from collected supernatants. These supernatants were analyzed with the human cytometric bead array (CBA) flex set (BD) according to manufacturer’s instructions. Briefly, a standard curve for each analyte using a four-parameter logistic regression analysis was created. This curve was based upon standards with fixed concentrations of each analyte and their corresponding median fluorescence intensities (MFIs). Then, MFIs of the various analytes within the samples were converted to concentrations (pg/mL). Samples were measured on the FACS Canto II (BD) and concentrations were determined with GraphPad Prism 5 (CA, USA).

Frequencies of IFN-γ-producing cells (spots/100,000 cells) following autologous, allogeneic, or PHA stimulation were measured with an Enzyme-Linked ImmunoSpot (ELISPOT) assay (U-CyTech, Utrecht, The Netherlands). During the day 1, the ELISPOT plate was coated with the antibody and incubated overnight. The same day cells were thawed and rested overnight. The following day, the assay plate was blocked using a blocking buffer and incubated for 1 h at 37°C. After the plate was washed with phosphate-buffered saline (PBS), cells were pipetted into wells and stimulated in triplicate, as described earlier, for 1 day. Thereafter, plates were washed first with PBS and then with PBS-Tween. Spots were made visible according to manufacturer’s instructions. Spots were analyzed using the ELISpot reader (Bioreader^®-600V, BIO-SYS GmbH, Karben, Germany).

Variables are presented as medians with interquartile ranges. Differences between paired samples (PB and LN from the same patient) were analyzed using the Wilcoxon signed-rank test. Differences between continuous variables from two independent groups were assessed with the Mann–Whitney U test. Differences between categorical variables were analyzed either with the Pearson’s chi-squared test or with the Fisher’s exact test depending on the expected values in any of the cells of a contingency table. Associations between rejection and the assessed parameters were analyzed using a binary logistic regression analysis. The significance level (p-value) was two-tailed and a value of α = 0.05 was used. Statistical analyses were performed using SPSS^® version 21.0 for Windows^® (SPSS Inc., IL, USA) and GraphPad Prism 5 (CA, USA). Figures were created with GraphPad Prism 5 (CA, USA).

Patient characteristics (n = 38) are shown in Table 1. Median patient age was 58 years and median donor age was 55 years. Most of the patients underwent an RT for the first time (82%), while 18% underwent a transplantation for the second time. Most common cause of ESRD was nephrosclerosis/atherosclerosis/hypertension (29%) followed by polycystic kidney disease (21%), together accounting for half of the cases. Overall, 24 (63%) patients received dialysis treatment prior to RT. Eleven of the 38 patients (29%) developed an EAR. Majority of the rejections were classified as cellular rejection (64%) and the remaining as antibody-mediated rejection. Median patient age was significantly younger within the rejectors compared with the non-rejectors (47 vs. 60, p < 0.019). All patients who developed an EAR were CMV-seropositive, while 63% of the non-rejectors was CMV-seropositive (p = 0.037). Median warm ischemia time was longer for the rejectors compared with the non-rejectors (23 min vs. 17 min, p = 0.045). Majority of the rejectors had dialysis prior to RT (91%), while 52% of the non-rejectors received this therapy (p = 0.030). Immunological risk factors like PRA score, HLA mismatches, or unrelated donor transplantation were not different between the two groups.

Figures 1 and 2 show the T-cell differentiation status of CD4⁺ and CD8⁺ T cells within the LNs of the rejectors and non-rejectors. Median frequency of CD4⁺ T cells prior to RT is significantly lower (p = 0.014) and median frequency of CD8⁺ T cells is significantly higher (p = 0.019) within the rejectors (Figures 1A,B). This led to a significantly lower CD4:CD8 ratio in LNs of patients with EAR (3.4 vs. 5.7, p = 0.027) (Figure 1C). Furthermore, median frequencies of CD4⁺ EM T cells and CD4⁺CD28^null T cells were significantly higher within the rejectors (p = 0.044 and p = 0.008, respectively) (Figures 2C,G). But overall, median percentage of CD4⁺CD28^null T cells remained low in both groups (0.8 vs. 1.5%). No other significant differences were observed for the other CD4⁺ or CD8⁺ T-cell subsets.

Assessment of T-cell differentiation status can be further refined by differential expression levels of CD27, CD57, and PD-1 in combination with CD28. However, this approach did not show any differences between the two groups (Table 2).

In summary, a lower CD4:CD8 ratio was associated with rejection.

The telomere length is indicative of T-cell replicative history, while CD31 expression on naïve T cells and TREC content are a measure of thymic output of naïve T cells. These parameters are strongly associated with the age and therefore a measure of the biological age of the T-cell system (31–36). Figure 3 demonstrates the RTL of T cells, RTE frequency, and the TREC content in LNs between the rejectors and non-rejectors. RTL of CD4⁺ and CD8⁺ T cells was similar between the two groups (Figures 3A,B). Also, no differences were observed for CD4⁺ and CD8⁺ RTEs between rejection and no rejection (Figures 3C,D). Finally, the TREC content was also not associated with development of rejection (Figure 3E).

Table 3 shows a univariate regression analysis of the significant clinical and immunological parameters with development of EAR. This univariate analysis showed that a higher CD4:CD8 ratio was associated with a lower risk for EAR (OR = 0.67, p = 0.039). Next to this, an older patient age and preemptive RT were also associated with a lower risk for rejection (OR = 0.93, p = 0.024; OR = 0.11, p = 0.046, respectively). Then these parameters were put into a multivariate regression analysis for a first model (Table 4). This showed that age was not associated with development of EAR. On the other hand, a higher CD4:CD8 ratio showed a tendency for a lower risk for rejection (OR = 0.63, p = 0.053), while preemptive RT still showed a significant association with development of EAR (OR = 0.06, p = 0.023). Taking the parameter “age” out of the equation showed a second model with significant associations (Table 4). A higher CD4:CD8 ratio and preemptive RT were both associated with a lower risk for EAR (OR = 0.58, p = 0.019; OR = 0.05, p = 0.012, respectively).

Thus, a lower CD4:CD8 ratio in the LN and receiving dialysis prior to transplantation increases the risk for the development of rejection after transplantation.

Median frequency of proliferating CD4⁺ T cells from the LN rose significantly from 2.4 to 45.0% (p = 0.002) upon allogeneic stimulation (Figure 4A), and that of CD4⁺ T cells from the PB increased from 2.1 to 39.3% (p = 0.004) (Figure 4B). Within the proliferating CD4⁺ T cells, majority of the cells were memory T cells (mainly CM) (Figures 4A,B). The CD28⁺ T-cell fraction was the main proliferating fraction within proliferating CD4⁺ T cells in PB and LN (>99.0% proliferation in both compartments) (Figures 4A,B).

The CD8⁺ T cells showed similar results compared with the CD4⁺ T cells. Median frequency of proliferating CD8⁺ T cells from the LN rose significantly from 1.4 to 22.9% (p = 0.004) upon allogeneic stimulation (Figure 4C) and median frequency of CD8⁺ T cells from PB increased from 0.6 to 29.9% (p = 0.002) (Figure 4D). CD8⁺ memory T cells comprised the largest fraction within proliferating CD8⁺ T cells in both LN and PB, with CM T cells also being the major contributor within this fraction (Figures 4C,D). The CD28⁺ T-cell fraction was again the main proliferating fraction within proliferating CD8⁺ T cells in LN and PB (Figures 4C,D).

Comparing the frequency of proliferating T cells upon alloreactive stimulation between LN and PB showed no differences except for CD8⁺CD28^null T cells within proliferating CD8⁺ T cells. Median frequency of these cells within LN amounted to 0.3%, while this was 2.9% within PB (p = 0.037), which is probably due to the higher presence of these cells within PB compared with LN.

Polyclonal stimulation with PHA induced >95% proliferation of CD4⁺ T cells and CD8⁺ T cells from both compartments, although this response was significantly higher in CD4⁺ and CD8⁺ T cells from LN (p = 0.037 for both subsets). The functional assays could not be performed in all patients as the yield of cells from the LN was not sufficient enough in some cases. For this reason, the number of patients with EAR included in this part of the study was too low (n = 3) for adequate comparison with patients without EAR.

In conclusion, the alloreactive potential is not different between T cells from LN and PB, but T cells from LN show a slightly higher proliferative potential in response to PHA.

Frequency of IFN-γ-producing LNMCs increased significantly after allogeneic stimulation (p = 0.034), while a similar tendency was seen for PBMCs (p = 0.083). However, median amount of spots remained low after allogeneic stimulation (3.5/1 × 10⁵ for the LN and 6/1 × 10⁵ for the PB). Furthermore, there were no significant differences between the two different compartments after allogeneic stimulation (Figure 5A). Stimulation with PHA showed a significantly higher IFN-γ secretion by PBMCs compared with T cells within LNMCs (256/1 × 10⁵ vs. 126/1 × 10⁵, p = 0.047).

After allogeneic stimulation of LNMCs and PBMCs for 6 days, concentrations of TNF-α, IFN-γ and granzyme B in the supernatants were significantly increased. LNMCs had a lower production of TNF-α (p = 0.065) (Figure 5B), IFN-γ, and granzyme B compared with PBMCs (p = 0.020 and p = 0.049, respectively) (Figures 5C,D). Interestingly, amount of IFN-γ spots after allogeneic stimulation for 1 day correlated significantly with production of IFN-γ by LNMCs as well as PBMCs after 6 days (ρ = 0.88, p = 0.001 and ρ = 0.64, p = 0.048, respectively).

Thus, production of TNF-α, IFN-γ, and granzyme B by LNMCs and PBMCs increases after allogeneic stimulation. However, generation of these products is lower by LNMCs than by PBMCs.