To determine the capacity of T cells to get activated under immunosuppression, we prospectively enrolled 22 patients who underwent kidney transplantation at the Lyon University Hospital between 2015 and 2016. Inclusion criteria were (i) first transplantation (kidney or kidney–pancreas), (ii) no anti-HLA antibody at the time of the transplantation. These 22 patients were compared to 9 healthy controls.

Whole blood samples were collected by venepuncture into heparin-containing vials, once in controls and before transplantation and at 3 months and 1 year after transplantation for transplanted patients.

Peripheral blood mononuclear cells (PBMCs) and plasma were isolated by Ficoll gradient centrifugation. Plasma was then centrifuged at 4,000 g for 10 min to remove platelets. PBMCs were plated 1 h in petri dishes to discard adherent cells (monocytes) and then 1 × 10⁶ non-adherent cells were cultured 24 h at 37°C in 5% CO₂ in 1 mL of patient’s own plasma (containing or not immunosuppressive drugs) in the presence or absence of human T-activator CD3/CD28 beads (Gibco Dynabeads^®). For IL21 staining, 1 µL of Brefeldin A (BD Bioscience) was added for the last 5 h of the culture. After 24 h of culture, Dynabeads were removed with a magnet and PBMCs were analyzed by flow cytometry as detailed below.

This study was carried out in accordance with French legislation on biomedical research. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol and the biocollection were authorized by the Ministry of Research and the Rhône-Alpes Regional Health Agency (#AC-2011-1375 and #AC-2016-2706).

Wild-type C57BL/6 (H-2^b) and BALB/c (H-2^d) mice were purchased from Charles River Laboratories (Saint Germain sur l’Arbresle, France).

C57BL/6-Tg(CAG-EGFP)1Osb/j (GFP) mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA).

MHC II knock out (AßKO) mice on C57BL/6 genetic background were provided by Dr. Benoist and Mathis (Boston, MA, USA) (14).

HLA A2 transgenic mice on C57BL/6 genetic background (15) were kindly provided by Dr Lemonnier (Paris, France).

RAG2 Knock Out (RAG2 KO) mice on C57BL/6 background came from CDTA (Cryopreservation Distribution Typage et Archivage animal, Orléans, France). Mice 8–15 weeks of age were used and maintained under specific pathogen-free conditions in our animal facility: Plateau de Biologie Expérimentale de la Souris (PBES, ENS Lyon, France).

This study was carried out in accordance with the French legislation on live animal experimentation. The protocol was approved both by local “CECCAPP” (http://www.sfr-biosciences.fr/ethique/experimentation-animale/ceccapp) and national AFiS ethical committees (#02870.01).

Seminal experimental studies addressing the role of helper T cells in rejection in vivo have relied on the treatment of wild-type (WT) mice with depleting anti-CD4 monoclonal antibodies (mAbs) (18, 19).

In the present study, we used IP administration of GK1.5, a rat IgG2b anti-murine CD4 mAb (20), commonly used to induce CD4+ T cell depletion in mice (21).

The human erythroleukemia cell line K562 (22) lacking expression of all MHC I and II molecules and the single antigen expressing cell Line (SAL) A2, a K562 cell line transfected with plasmid coding for HLA A2 and selection genes for the resistance to neomycin and ampicillin (23) were kindly provided by Dr. Doxiadis (Leiden, The Netherlands).

K562 cells were cultured in RPMI-1640 (Invitrogen) complemented with fetal calf serum 10% (Dutscher), l-Glutamine 2 mM (Invitrogen), penicillin 100 U/mL, streptomycin 100 µM, and HEPES 25 mM (Invitrogen).

Single antigen expressing cell line A2 cells were cultured in the same medium as K562 complemented with G418, a neomycin derivative (Invitrogen) at the concentration of 1.75 mg/mL.

After a 24 h culture, human PBMCs were stained 30 min at room temperature in the dark with the relevant fluorescent antibodies directed against CD3 (UCHT1), CD4 (SK3), CD8 (SK1), CD25 (2 A3), CXCR5 (RF8B2), CD40L (TRAP1), IL21 (3A3-N2.1), ICOS (ISA-3); all from BD Biosciences except ISA-3 from ebioscience. Cells were then stained with and a viability dye LIVE/DEAD Aqua (Invitrogen) according to the manufacturer’s instructions and fixed with Cytofix/Cytoperm^® fixation/permeabilization kit (BD Biosciences) before analysis.

A FACS ARIA II flow cytometer was used for flow cytometry. Data were analysed with BD FACS Diva software (BD Biosciences).

Murine single-cell suspensions from spleen, lymph nodes, thymus, or blood were incubated with a blocking anti-mouse Fc receptor antibody (2.4G2, home-made hybrydoma). Cells were then incubated at 4°C with relevant fluorescent antibodies: CD3 (145-2C11), CD4 (RM4-4), CD8 (53-6.7) and CD19 (1D3) (all from BD Biosciences). Before analysis by flow cytometry, DAPI (4′,6-diamidino-2-phenylindole dihydrochloride, Sigma-Aldrich) was added to the cell suspension to exclude dead cells. Samples acquisitions were made on a BD LSR II flow cytometer (BD Biosciences). Data were analyzed with FlowJo software (Tree Star).

For isolation of B lymphocytes, splenocytes from immunized mice were stained with phycoerythrin (PE)-conjugated antibodies against CD3ε (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD11b (M1/70), CD11c (HL3), NK-1.1 (PK136), Ter119/Erythroid cells (TER-119), CD117 (2B8), and PerCP-Cy5.5 conjugated antibody against B220 (RA3-6B2) (all from BD Biosciences). After staining, cells were negatively separated by LD magnetic columns with anti-PE Microbeads labeling (Miltenyi Biotec). The separated cell suspensions underwent cell sorting by using a BD FACS Aria cell sorter (BD Biosciences) to collect PE negative and PerCP-Cy5.5 positive cells. 5 × 10⁶ purified B lymphocytes were transferred to RAG2 KO mice by intravenous injection.

1 × 10⁵ SAL A2 or K562 (negative control) cells were incubated with mice sera diluted at 1/1,000 for 30 min at 4°C. After washing, cells were stained with PE-conjugated anti-light chain antibody (187.1) (BD Biosciences) and analyzed by flow cytometry. To rule out the possibility to miss a low titer of anti-HLA A2 antibodies, all negative sera were systematically screened again at a 1/100 dilution. The titer of anti-HLA A2 antibodies at each time point (dx) was calculated with the following formula: normalised DSA titer=MFISAL(dx)/MFIK562(dx)MFISAL(d0)/MFIK562(d0)

The normalized DSA titer, therefore, reflects the fold increases of specific signal over the baseline (measured before transplantation).

Single-cell suspensions were stained with PE-conjugated anti-B220 (RA3-6B2) (BD Biosciences). B cells were positively separated by passing through LS magnetic column with anti-PE microbeads labeling. After separation, cells underwent in vitro activation with a mixture of the TLR7/8 ligand R848 at 1 µg/mL and recombinant mouse IL-2 at 10 ng/mL (Mabtech) for 72 h. Then, cells were cultured for 24 h in the 96-well plates precoated with capture anti-IgG antibody (ELISpot^PLUS mouse IgG kit, Mabtech). Detection of spots was performed by adding either anti-IgG biotinylated detection antibody (Mabtech) or biotin-labeled HLA-A*02:01 Pentamer (ProImmune) for 2 h, followed by incubation with streptavidin-ALP (Mabtech) for 1 h and development by BCIP/NBT-plus substrate (Mabtech).

For some experiments, mice were immunized intraperitoneally with 75 µg NP-KLH mixed with 100 µL Inject Alum Adjuvant (Thermo Scientific, Courtaboeuf, France) or 200 µg NP-Dextran. Preimmune sera and sera obtained once a week after immunization were tested for IgM and IgG anti-NP antibodies. Maxisorp plates (Nunc) were coated with NP 23-conjugated BSA. Serially diluted serum samples were added for 1 h 30 at room temperature. Anti-NP IgM and IgG Abs were detected using alkaline phosphatase conjugated goat anti-mouse IgM or IgG Abs (1/2,000 dilution) followed by phosphatase substrate (Sigma-Aldrich). The plates were then read at 405 nm/490 nm with an automatic reader (Zeiss VERSAmax). OD was converted to concentration based on standard curves with sera from C57BL/6 mice immunized with NP-KLH using a four-parameter logistic equation (Softmax Pro 5.3 software; Molecular Devices).

Statistical significance of differences was tested with chi-square test for proportions and non-parametric tests for continuous variables: Mann–Whitney or Kruskal–Wallis with Dunn’s multiple comparisons for unpaired dataset and Friedman with Dunn’s multiple comparisons for paired dataset.

Skin graft survivals were compared using the log-rank test.

The differences between the groups were considered statistically significant for p < 0.05 and were reported as asterisk symbols (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).

AβKO mice, which are genetically modified C57BL/6 mice devoid of CD4+ T cells were used as recipients. AβKO mice lack major histocompatibility complex class II (MHCII) molecules (14). As a result, positive selection of CD4+ T cells can not occur in the thymus and AβKO mice show near-complete elimination of CD4+ T lymphocytes in the periphery, the lymph nodes, and the spleen (Figure 1A). The few remaining detectable CD4+ T cells are anyhow unable to provide efficient help to B cells because the latter do not express MHCII. This was demonstrated by the complete abrogation of the IgG response to a NP-KLH, a model of thymo-dependent antigen (24) (Figure 1B). Importantly, this lack of IgG response in AβKO mice was not due to an abnormal B cell compartment since (i) B-lymphocytes occur in normal numbers and are capable of terminal differentiation to plasma cells in these animals (14), and (ii) we showed that AβKO mice indeed generated NP-specific IgM in response to vaccination with NP-dextran (Figure 1C), a model of thymo-independent antigen (25).

To simplify the monitoring of donor-specific antibody (DSA) response in recipient mice we used donors with the same genetic background (C57BL/6, H2^b) except for the transgenic expression of the human MHC I molecule HLA-A2 [kind gift from Dr Lemonnier (15); Figure 1D].

In this model, the DSA response, which is exclusively directed against the HLA-A2 molecule was easily monitored using single antigen-expressing (SAL-A2) cell line [kind gift from Doxiadis (23)] and the sensitive and specific flow cross match assay (Figures 1E,F), a technique routinely used for the follow-up of transplant recipients in the clinic (26).

Because AβKO mice have a normal CD8+ T cell compartment (Figure 1A), they efficiently rejected an allogeneic A2 skin graft with only a slight delay as compared with C57BL/6 WT controls (respective mean survival 14.75 ± 1.5 vs 8.75 ± 1.5 days, Log Rank: p = 0.0058; Figure 2A).

Donor-specific antibody response was monitored in naive recipients grafted with an A2 skin. Anti HLA-A2 alloantibodies appeared in the serum of WT recipients between day 7 and 14 posttransplantation, peaked at day 21, and decreased slowly thereafter with the elimination of alloantigens due to the rejection process (Figure 2B). No DSA could be detected in the serum of AβKO.

These results confirm the conclusion of Steele et al. (27) who reported that CD4+ T cells are essential helper cells for B cell alloantibody production in a similar murine model of MHC-disparate skin graft rejection. However, in the study by Ochsenbein et al., dependence on T cell help went decreasing when increasing amounts of antigen were recruited to the spleen (13). One possibility to explain the lack of DSA response in AβKO recipients grafted with A2 skin could be that this procedure provides too few alloantigens drained to the spleen of recipients (in contrast with transplantation, no vascular anastomosis is performed between the circulations of donor and recipient during grafting). To test this hypothesis, C57BL/6 donor mice transgenic for the green fluorescent protein (GFP+) were used as donor of skin graft or heart transplant (Figure 2C). The proportion (Figure 2D) and absolute number (Figure 2E) of GFP positive cells were quantified by flow cytometry at day 2 post-procedure in the draining lymph node, a distant non-draining lymph node, and the spleen of recipients. After skin grafting, only few GFP+ cells were detected, exclusively in the draining lymph node of recipient (Figures 2D,E). About 100 times more GFP+ cells were found in the draining lymph node of heart recipients. In addition, GFP+ cells were also detected in the spleen and the non-draining lymph node after heart transplantation (Figure 2D), demonstrating that alloantigens are drained trough recipient’s circulation after transplantation but not after grafting. In fact, when absolute numbers were considered, spleen appeared as the main site of allosensitization after transplantation (Figure 2E).

Despite these differences in alloantigen dose and localization in secondary lymphoid organs of recipients, we could not detect any DSA generation after A2 heart transplantation in AβKO recipients (Figure 2F). We conclude from this set of experiments that DSA generation in naive recipients requires CD4+ T cell help.

In contrast with laboratory mice, the immune system of which is largely naïve (28), transplanted patients have an immune memory, including memory B cells (29). Interestingly, some studies have reported that ability of virus-specific memory B cells to secrete IgG is independent of cognate or bystander T cell help (30).

Transplantation with an A2 but not a BALB/c heart leads to the generation of anti-A2 specific memory B cells that could be readily quantified in recipients’ spleen with ELISpot assay (Figure 3A). To determine whether the rechallenge of allospecific memory B cells can lead to the generation of DSA in the absence of CD4+ T cells, we passively transferred A2- or BALB/c-specific memory B cells to C57BL/6 RAG2 KO recipients (Figure 3B). C57BL/6 RAG2 KO mice lack the recombinase necessary for TCR and BCR rearrangements and are, therefore, devoid of T and B cells (Figure 3C). In addition to the high degree of purity achieved when isolating B cells (99.83 ± 0.21%, Figure 3D), we ensured that no CD4+ T cell remained in C57BL/6 RAG2 KO post-cell transfer by administering anti-CD4 mAb (Figure 3B). As expected, B cells (but neither CD4+ nor CD8+ T cells) could be detected in the circulation of transferred C57BL/6 RAG2 KO animals up to the end of the follow-up period (Figure 3C). Not only were B cells detectable in transferred C57BL/6 RAG2 KO mice, but A2-specific memory B cells were able to differentiate into DSA-producing plasma cells, as demonstrated by ELISpot assays conducted with splenocytes harvested 100 days post-transfer (Figure 3E).

However, and in striking contradiction with our hypothesis, the rechallenge of allospecific memory B cells with A2 heart did not lead to the generation of DSA in the absence of CD4+ T cells (Figure 3F). The mandatory nature of CD4+ T cell help for DSA generation in sensitized recipients was further confirmed by demonstrating that co-transfer of purified CD4+ T cells with memory B cells in C57BL/6 RAG2 KO mice was sufficient to restore alloantibody response after rechallenge (Figures S1C,D in Supplementary Material).

Based on our previous results, it seems clear that DSA generation, both in naive and sensitized recipients, requires help from CD4+ T cells. The fact that, despite therapeutic immunosuppression, 10–20% of transplant recipients develop de novo DSA within 5 years posttransplantation (3–5), suggests that immunosuppressive drugs do not efficiently block CD4+ T cell help to B cells.

To test this hypothesis, we prospectively monitored circulating T cells in 22 immunosuppressed renal recipients, 3 months (3 M) and 12 months (12 M) posttransplantation (Figure 4A). Clinical characteristics of the cohort are presented Table 1 and the trough levels of maintenance immunosuppressive drugs are shown Figure 4B. Immunosuppressed patients were compared to 9 age- and sex-matched healthy controls. Indeed, comparing the evolution of T cell parameters before versus after transplantation for the same patients would not have allowed to disentangle the impact of immunosuppressive drugs from that of the correction of end stage renal failure, which has a major impact on the immune system (31).

To test whether maintenance immunosuppressive regimen is able to block T cells activation, PBMCs of patients were cultured with anti-CD3/CD28 microbeads in the presence of patient’s own plasma (Figure 4A). Surface expression of the α chain of IL-2 receptor (CD25) was used to monitor activation of CD8+ and CD4+ T cells (32). As expected, in vitro stimulation of T cells from healthy controls led to significant upregulation of CD25 expression on both CD4+ and CD8+ subsets (Figure 4C). Interestingly, while the presence of plasma containing immunosuppressive drugs significantly reduced activation-induced upregulation of CD25 on CD8+ T cells, it had no significant impact on CD4+ T cells (Figure 4C).

To further document the lack of impact of maintenance immunosuppression on thymo-dependent humoral immune response, we focused our analysis on T follicular helper cells (Tfh), a CD4+ T cell subset specialized for providing help to B cells (33, 34). Recent studies have indeed demonstrated that CD4+CXCR5+ human blood T cells represent the circulating compartment of Tfh (35), and we confirmed that CD3+CD4+CXCR5+Tfh could be detected in the circulation of both controls and transplanted patients (Figure 4A).

T follicular helper cells provide help to B cells through a variety of molecules that are either expressed on their surface: CD40L (36) and ICOS (37), or are secreted: IL-21, a type I cytokine recognized as the most potent driver of B cell terminal differentiation (38, 39). In vitro activation with anti-CD3/CD28 microbeads promoted the expression of CD40L (Figure 4D) and ICOS (Figure 4E) by the Tfh. As expected, triple maintenance immunosuppression significantly decreased the level of expression of both CD40L and ICOS on activated Tfh (Figures 4D,E). However, the impact of triple maintenance immunosuppression was incomplete because the drug combination failed to reduce the percentage of Tfh that expressed CD40L after stimulation (Figure 4D) or the production of IL21 by Tfh (Figure 4F).

Transplanted patients often receive induction therapy at the time of transplantation to prevent acute rejection during the early posttransplantation period. Induction immunosuppressive agents used in our center are either the lymphocyte-depleting agent rabbit antithymocyte globulin (ATG) or the anti-IL2 receptor mAb basiliximab, which is non-depleting.

Induction with ATG induced a 90% drop in the number of cTfh at 3 months (Figure 5A). Beyond this time point, the number of cTfh started to increase slowly and was no longer significantly different from baseline at 1 year posttransplantation (Figure 5A). As expected, induction with the non-depleting agent, basiliximab was not followed by any reduction in cTfh number. At the contrary, cTfh number was significantly increased (+36%) at 1 year posttransplantation in patients that received basiliximab (Figure 5A).

Our previous results indicate that while maintenance immunosuppression does not completely block Tfh functions, a prolonged depletion of cTfh can be achieved following induction with a depleting agent. We, therefore, analyzed the impact of such prolonged depletion in CD4+ T cells on naive and memory experimental DSA responses.

GK1.5 (20) is a rat IgG2b anti-murine CD4 mAb commonly used to induce CD4+ T cell depletion in mice (21). Using repeated IV administrations of different doses of GK1.5, we were able to obtain various levels of CD4+ T cell depletion (Figure 5B). Due to (i) the large number of animals to be analyzed and (ii) the complexity of microsurgical models of transplantation (sequential transplantation of two hearts in the cervical area of the same mice is unfeasible, making analysis of memory response impossible), we used a simplified model of allosensitisation (presented in Figure S1 in Supplementary Material). DSA responses of C57BL/6 WT mice were induced by IV injection of purified splenocytes from HLA-A2 transgenic mice (Figure 5C). CD4+ T cell depletion did not change the kinetic of DSA generation but resulted in significant reduction of DSA titers both in naive and memory responses (Figure 5D). Importantly, DSA titers strongly correlated with the number of remaining CD4+ T cells of recipient mice in an exponential regression model (r² for naive and memory responses, respectively, 0.86 and 0.84, Figure 5E).

Recovery of peripheral T-cell counts, including cTfh occurred gradually after cessation of ATG treatment (Figure 5A). However, beyond the quantitative impact of ATG, several studies have documented a long-term qualitative influence of the drug on T cell compartment (40). To determine if ATG induction could have long-term effects on Tfh function, we compared the profile of cTfh of renal recipients that received ATG or basiliximab as induction. The two groups did not differ regarding their clinical characteristics (Table 1) or maintenance regimen (Table 1; Figure 6A). Although cTfh functions of the two groups were similar before transplantation (Figures 6B,C), cTfh of renal recipients that received ATG exhibited a significantly reduced activation-induced upregulation of CD40L (Figure 6B) and a similar trend was observed for ICOS (Figure 6C).

These results suggest that ATG induction could reduce the incidence of de novo DSA after renal transplantation. We were, however, unable to test this hypothesis in this study because only three patients of the 22 enrolled, have developed de novo DSA over the follow-up period (2 in the group ATG induction, 1 in the group basiliximab induction).