Both preexisting and de novo donor-specific anti-HLA antibodies (DSAs) are associated with chronic antibody-mediated rejection (ABMR), a leading cause of renal allograft loss (1–6). However, biological events associated with the appearance of DSAs are poorly characterized, and strategic options to block these events have limited efficacy because they are non-specific, administered too late or inefficient at targeting the source of antibodies (7, 8). A better understanding of the mechanisms leading to the formation, development, persistence, and action of DSAs is needed to guide the development of novel strategies to control DSAs and improve transplantation outcomes.

As a trigger of B cell antibody production, T follicular helper cells (Tfhs) are instrumental in promoting DSA appearance (9, 10). Tfhs are essential in the formation of germinal centers and development of an immunological memory response (11, 12). After interacting with B cells, Tfhs emerge from the germinal centers to become circulating Tfhs (cTfhs) (13, 14). Because of limited access to secondary lymphoid organs in humans, the existence of these cTfhs, which share the same functional capacities of germinal center, offers interesting opportunities for study and access. cTfhs can induce in vitro B lymphocyte differentiation (13, 14) and migrate into grafts where they can contribute to tertiary lymphoid organs that are associated with rejection (10, 15). Different subsets of cTfhs have been described with specific functions and phenotypes, with differential expression markers including CXCR3, inducible T cell costimulator (ICOS) and programmed cell death protein 1 (PD1) (16). Notably, PD1 has been associated as an activation marker of cTfhs and ICOS was found to be needed for cTfh homing and functions into germinal centers (14, 17). The expression of ICOS was also associated with expression of Ki67 showing cTfhs in active cell cycle (18). The use of CXCR3 distinguished cTfhs able to promote naïve B cell differentiation and immunoglobulins production, named Th2 (CXCR3⁻CCR6⁻) and Th17 (CXCR3⁻CCR6⁺), from CXCR3⁺CCR6⁻ Th1 cells unable to help naïve B cells (13). The link between cTFhs and antibody production has been clearly demonstrated; CXCR5⁺PD1⁺CXCR3⁻ cTfhs are correlated with anti-HIV antibody development (14), whereas CXCR5⁺CXCR3⁺ICOS⁺ cTfhs are associated with the development of antibodies after seasonal influenza vaccination by providing help to memory B cells (18). Based on these findings, these cTfh subsets with an activated phenotype may be instrumental in antibody production after renal transplantation.

Several studies in animal models suggest a beneficial effect of blocking the differentiation of cTfhs in transplantation (9, 19–23). In Humans, recent reports have shown an association of Tfhs with anti-HLA antibodies and/or DSA (10, 24–29). Interestingly, in accordance with these data, we recently reported reduced proportions of activated CXCR5⁺PD1⁺, CXCR5⁺PD1⁺ICOS⁺ and highly functional CXCR5⁺PD1⁺CXCR3⁻ cTfh subsets in blood from tolerant patients who stopped all immunosuppressive treatment while maintaining a functioning graft (25). This cTfh defect was linked to a low incidence of postgraft de novo donor-specific antibody (dnDSA) immunization (25). These data suggested that cTfhs may be present long before DSA appearance but have not being evidenced yet. Moreover, cTfhs can migrate and their presence in the graft is clearly associated with graft dysfunction, as attested in biopsies from patients experiencing acute T cell-mediated rejection, ABMR with anti-HLA-II DSAs and with mixed T and B cell-mediated rejection (10, 15, 29).

Examining their presence/clearance in blood from transplanted patients, particularly activated cTfh subsets linked to antibody production, may thus predict the genesis of antibody-mediated rejection as suggested in mouse model (23). Thus, herein, we measured the frequency of total CD4⁺CD45RA⁻CXCR5⁺ cTfhs and activated CXCR5⁺PD1⁺, CXCR5⁺PD1⁺ICOS⁺ and CXCR5⁺PD1⁺CXCR3⁻ subsets at the time of transplantation (day 0) and after 1 year (M12) in peripheral blood mononuclear cells (PBMCs) from a prospective cohort of 237 patients without DSA before renal transplantation and without any occurrence of dnDSA during the first year post-transplantation. By measuring cTfh levels before dnDSA appearance, in a large and prospective cohort of transplanted patients with a median follow-up of more than 7 years, we reported a significant association between the activated CXCR5⁺PD1⁺ICOS⁺ cTfh subset from DSA-free patients at M12 after transplantation and the appearance of dnDSA immunization after the first year post-transplantation. We also showed that depleting induction treatment was associated with a relative increase of activated cTfhs suggesting that current treatments did not provide efficient prevention of Tfh activation.

The occurrence of dnDSA after transplantation is associated with an increase in the prevalence of ABMR and the risk of graft loss (1–6). Thus, monitoring dnDSA after transplantation is instrumental and recognized as a standard of care for renal transplant recipients. Tfhs are the main cells orchestrating DSA production by aiding B cells through different molecules, such as ICOS (32), CD40L and PD1, as well as various soluble factors (Il-4, Il-6, and Il-21) (12, 13, 33, 34). Tfhs are composed of phenotypically and functionally different subsets based on the expression of these molecules and their role has been demonstrated in several pathologies in humans (14, 18, 35).

Blood cTfhs share similar functional properties with classical Tfhs from the germinal center (13). In transplantation, cTfhs are associated with renal transplant outcomes (10, 24–27, 29). Specifically, higher levels of cTfhs (CD4⁺CXCR5⁺) were observed at diagnosis in patients who experienced ABMR (24), ABMR with anti-HLA class II DSA (29), and in patients with preexisting DSA at 3 months post-transplantation (10). A higher absolute number of cTfhs before transplantation was associated with acute rejection experiences during the first year of transplantation (27). Interestingly, in a cohort of tolerant patients, we observed a decrease in activated cTfh subsets (CXCR5⁺PD1⁺, CXCR5⁺PD1⁺ICOS⁺, and CXCR5⁺PD1⁺CXCR3⁻) without affecting the absolute number of cTfhs (25). In these previous studies, cTh levels were measured at or after diagnostics, for chronic rejection (10, 24, 29) or tolerance (25). However, no study has yet demonstrated cTfh levels as predictive of DSA appearance in the long term in Human. To answer this point, we measured cTfhs in a large and prospective cohort of 237 DSA-free patients at day 0 and M12 with a follow-up of 89.6 months after transplantation. We reported that the M12 CXCR5⁺PD1⁺ICOS⁺ cTfh level was significantly and independently associated with dnDSA appearance after 1 year post-transplantation. DSA-free patients with lower levels of CXCR5⁺PD1⁺ICOS⁺ cTfhs at M12 had a higher risk of developing dnDSA after the first year post-transplantation when B cell modifications could not be detected. Furthermore, in our prospective study, patients who developed dnDSA over 1 year after transplantation exhibited worse allograft survival than patients who did not develop dnDSA (p < 0.0001), in accordance with other reports evidencing dnDSA appearance was a negative risk factor to allograft survival (1–6).

Cano-Romero's group recently reported that patients who develop de novo anti-HLA antibodies during the first year post-transplantation display an increased number of PD1⁺ cTfhs up to 6 months after transplantation (27). The authors analyzed de novo anti-HLA antibody appearance during the first year post-transplantation, whereas we focused on dnDSA occurrence after the first year of transplantation in patients who were DSA-free in this first year. In a murine model of skin transplantation, a rapid increase of donor-specific CXCR5⁺PD1⁺ICOS⁺ cTfhs in blood preceded the appearance of DSA, followed by a contraction phase (23). By contrast, we did not observe an increase of these cTfhs but a decrease level associated with dnDSA which may be explained by different kinetics between mice and human (days compared to months or years). Our different endpoints and results suggest different cell kinetics for cTfhs, particularly activated CXCR5⁺PD1⁺ICOS⁺ cTfhs, which may have expanded the first months after transplantation according Cano-Romero et al. results, and then migrated later after transplantation in the graft and decreased in peripheral blood (27). In line with this hypothesis, Shi et al. evidenced an increase of CD4⁺CXCR5⁺ cTfhs in peripheral blood at diagnosis of ABMR with a decrease expression of PD1 that we could linked to the decrease frequency of activated cTfhs we observed (24). Indeed, ICOS is essential to the localization of CD4⁺ T cells within follicles and to support a germinal center response (17) and cTfhs have been evidenced to migrate toward inflamed organs such as in lungs infected with influenza virus (36). We hypothesized the decreased frequency of activated cTfhs suggests their migration toward draining lymph nodes to support antibody production (19) or migration into the graft where local inflammation is taking place to participate in the formation of tertiary organs and establishing a local memory compartment (10, 16, 29, 37). Indeed, according literature, cTfhs are likely to be present at diagnosis in biopsies from patients experiencing allograft rejection (10, 15, 20, 29). Liarski et al. demonstrated that CD4⁺PD1⁺ICOS⁺ cells were found in 50% of renal biopsies with mixed T and B cell-mediated rejection and expressed high levels of IL21 mRNA, which is a key cytokine to promote B cell maturation (15). The presence of CD3⁺CXCR5⁺ Tfhs have been evidenced in biopsies from patients exhibiting anti-class II DSA and ABMR as well as the presence of Bcl-6-expressing Tfhs on the T-B cell border of the follicular-like structures in biopsies from acute TCMR (10, 29). We also evidenced the presence of CD4⁺PD1⁺ T cells co-expressing ICOS and IL-21 in a model of renal transplantation (20). However, our current analysis was focused on peripheral blood as a non-invasive compartment which also allows repeated measures for immune monitoring. Evidences of cTfh roles in dnDSA formation and allograft rejection would require further investigations, notably in a longitudinal study with blood and matched biopsies.

In our cohort, ATG treatment was associated with a reduction in the total CXCR5⁺ cTfh number at M12. ATG induces depletion of both CD4⁺ and CD8 T⁺ cells (38), and cTfh cells represent approximately 15% of total circulating CD4⁺ T cells in healthy individuals. Thus, unsurprisingly, induction treatment and standard immunosuppressive therapies decreased cTfh levels as previously reported (27, 28, 39). Our data are also consistent with results from Thaunat's group who reported prolonged cTfh depletion after ATG induction therapy and reduced CD40L and ICOS expression on activated cTfhs (28). In our cohort, this decreased frequency of total cTfhs was associated with increased frequencies of activated cTfh subsets (CXCR5⁺PD1⁺, CXCR5⁺PD1⁺ICOS⁺, and CXCR5⁺PD1⁺CXCR3⁻ cTfhs) in CNI-treated recipients and ATG induction therapy, as well as higher expression of cTfh activation marker PD1 at M12. The frequency of activated cTfh subsets is also increased by immunizing events, including previous blood transfusions and/or pregnancies, and the presence of class II non-DSA anti-HLA antibodies. These data, in accordance with previous reports, suggest that pre-sensibilization increases the pool of activated cTfhs (10, 27, 40) and is consistent with the lack of effect of standard immunosuppression on activated Tfhs (23, 28). These results are also in line with reports exhibiting that T cells were unequally targeted by immunosuppressive treatments, memory T cells exhibiting a relative increase frequency after ATG depletion (41). In addition, we were not able to identify significant difference in cTfh distribution between patients treated with corticoid treatment or not. These data clearly suggest that current immunosuppressive therapies cannot (at least not efficiently) target activated Tfhs, which are potentially harmful for the graft. T cells that are refractory to depletion likely continue to aid B cells. As a limitation of our study, cTfh phenotype could be altered by cell freezing, however, we did not observe any impact of storage time on viability percentages or total cTfh frequency (Supplementary Figure 3). Due to unbalanced administration of other immunosuppressive treatments, we could not analyze their effects on cTfh levels, limiting our ability to generalize our conclusion to all immunosuppressive treatments.

In conclusion, our study shows that one-year CXCR5⁺PD1⁺ICOS⁺ cTfh level was significantly and independently associated with dnDSA appearance after 1 year post-transplantation, before dnDSA appearance, and that current immunosuppressive therapies are unable to efficiently target the pool of activated cTfhs. DSA-free patients with lower levels of M12 CXCR5⁺PD1⁺ICOS⁺ cTfhs had a higher risk of developing dnDSA after the first year post-transplantation. Tfhs are involved in all types of rejection; beyond mixed or antibody-mediated effects alone, the maintenance of their function under immunosuppression allows B cells to exert their cytotoxic and antigen-presenting functions. Our results encourage the development of immunosuppressive strategies targeting Tfhs and activated subsets, specifically CXCR5⁺PD1⁺ICOS⁺ cTfhs, to prevent dnDSA appearance and the risk of CAMR. Clearly, these data also raise the need of following Tfh to predict DSA occurrence in patients.

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

This non-interventional research project involved data from the DIVAT cohort (www.divat.fr) and samples from the CENTAURE biocollection (www.fondation-centaure.org/) declared since 13/08/2008 to the Ministry of Research (N° PFS08-017) and hosted by the Centre de Ressource Biologique (CRB) of Nantes University Hospital. Samples recorded in CRB software are in connection with the clinical data of the DIVAT database, in line with the good practice recommendations of the University Hospital of Nantes [approved by the CNIL (DR-2025-087 N° 914184, 15/02/2015) and the French Ministry of Higher Education and Research (file 13.334-cohort DIVAT RC12_0452)]. For each patient, written consent was obtained to use the clinical and laboratory data. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul.

ACh, MG, and SB designed the study. MC, FD, and SL carried out the experiments. RD, MC, FD, CK, and ACe analyzed the data. RD, MC, SC, ND, MG, and SB drafted and revised the paper. All authors approved the final version of the manuscript.