Combining CFSE-dilution as a read out with CD40-B cells as stimulator cells, we developed a robust assay that detects human T cells with direct allospecificity at frequencies comparable to those predicted by animal experiments and 1- to 3-logs higher than PF detected with conventional LDA or ElLISPOT 
. This CFSE-MLR assay is able to discriminate between CD4 and CD8 PF and detects both naïve and memory T-cell responses to allo-antigens in the same culture. By application of this assay we found that, even with immunosuppressive therapy, numbers of circulating CD4+
T cells with direct donor-specificity increase immediately after LTx, indicating that both subsets are primed by the graft, followed by a gradual decrease within the first year. Importantly, after correction of changes in responses to 3rd
party allo-antigens, no specific loss of donor-specific T cells was observed, showing that CD4+
T cells with direct donor-specificity persist in the circulation up to at least 1 year after LTx. However, a non-specific decrease in frequencies of circulating allo-reactive CD4+ and CD8+ T cells to values slightly below pre-transplant values was observed 1 year after LTx, a well as a non-specific decrease in their capacity to produce IFN-γ.
The first improvement contributing to the robustness of the described assay is utilizing CD40-B cells as stimulators, resulting in 3- to 5-fold higher PF compared to stimulations with splenocytes. CD40-engagement induces B-cell differentiation into professional APC with uniform expression of co-stimulatory molecules, that are able to prime both memory and naïve T-cell reponses 
,  
. Contrary, splenocytes contain only about 50% APC, of which a minority express co-stimulatory molecules. CD40-B cells can be expanded from thawed splenocytes or PBMC, even when these have a low viability, and their expansion enables repeated measurements of donor-specific T-cell reactivity in cases of limited supply of donor cells. Importantly, the absence of T cells within CD40-B cell preparations precludes that donor-derived T cells are included in CFSE-profiles of responder T cells, which would lead to wrong PF calculations.
The second improvement is the use of CFSE-dilution in combination with software that calculates PF as a read-out technique. Detection of responding cells by this technique is not dependent on the number of progeny cells in each individual culture, which in LDA may lead to underestimation of PF when small clones of progeny cells remain undetected 
. Moreover, estimated PF are not negatively influenced by “preparation-induced cell death” of responder T cells at the beginning of the cultures, because PF are calculated from the sum of precursors present at the end of the cultures instead from cell input at the start of the culture. We showed that allogeneic stimulations are prone to significant loss of responder T cells during the first 2 days of culture. Since dying cells do not respond, techniques calculating PF from numbers of cells present at the start of the culture, like LDA and ELISPOT, underestimate actual PF 
. Conversely, the CFSE-MLR may overestimate PF if more non-proliferating cells compared to proliferating cells die during culture. It is difficult to analyze whether non-proliferating cells die, but the appearance of a plateau phase of PF between day 5 and 6 of culture indicates that at least during this last phase of the cultures no preferential death of non-proliferating T cells occurs.
ELISPOT also underestimates PF by ignoring allo-reactive T cells that do not secrete the particular cytokine detected. Indeed, we observed that not all T-cells that responded to allogeneic CD40-B cells by cell division (as determined by CFSE-dilution) produced IFN-γ. In addition, our data show that the numbers of IFN-γ producing T cells detected in ELISPOT are considerably lower compared to those detected by flowcytometry in CFSE-MLR, demonstrating the lower sensitivity of the ELISPOT-technique. In contrast to short-term ELISPOT assays which detect only the rapidly reacting allo-reactive memory T cells 
, the CFSE-MLR detects allo-reactive precursors in memory and naïve T cells simultaneously. Because the allo-reactive repertoire contains both naive and memory T cells 
, we judge that simultaneous quantification of both subsets is an advantage of the current technique compared to ELISPOT. Together, these differences may explain why published PF of donor-specific T cells in human organ transplant recipients detected by LDA 
or ELISPOT 
are significantly lower compared to those observed in this study with the CFSE-MLR. Importantly, the use of CD40-B cells as stimulators to detect allogeneic T-cell responses 
, and CFSE-dilution as a technique to measure the proliferative response of T cells to allo-antigens 
have both been described, but to our best knowledge these techniques have never been combined.
It is generally assumed that recipient T-cell responses against directly presented donor allo-antigen peak shortly after transplantation 
, but this has never been formally proven in humans. So far, the only other study that quantified donor-specific T-cell responses longitudinally after human organ transplantation 
actually found a nadir in frequencies of donor-specific T cells at 1 week after kidney transplantation using IFN-γ ELISPOT. We observed, after correction for changes in responses to 3rd
party allo-antigens, a significant, but modest, increase of circulating donor-specific CD4+
T-cell numbers immediately after LTx, followed by a decrease to levels equal to numbers of T cells responding to 3rd
party allo-antigens within the first year. This increase was not due to variations in immunosuppression, because it was corrected for variations in 3rd
party responses. Moreover, the highest blood trough concentrations of the calcineurin inhibitors CsA and Tacrolimus were observed in these patients at 1 week after LTx, after which they gradually decreased (data not shown), indicating that the increase of numbers of donor-specific T cells during the first week after LTx occurred despite high levels of immunosuppressive drugs, and that the decrease afterwards was not the consequence of increasing levels of immunosuppressive drugs. The observed increase in T-cell PF reacting to 3rd
party allo-antigens, which was smaller compared to the increase in donor-specific T-cell PF, may be explained by partial overlap in HLA between donor and 3rd
party stimulator cells spleen CD40-B cells. Due to limited availability of banked splenocytes, complete HLA mismatching between donor and 3rd
party stimulators was not always possible. Our results are congruent with recent observations in mice showing a significant increase in direct pathway donor-specific T-cells shortly after transplantation 
, which is probably related to migration of donor-derived dendritic cells from the graft into the recipient 
With LDA, it has repeatedly been shown that after organ transplantation in humans, numbers of circulating donor-specific T cells decrease below pre-transplant values and below numbers of T cells responding to 3rd
party allo-antigens 
. Donor-specific hypo-responsiveness was already observed at 1 month after LTx 
. This phenomenon was attributed to induction of anergy in donor-specific T cells 
or to suppression exerted by CD4+
Foxp3+ regulatory T cells 
. Of notice, donor-specific hypo-responsiveness in cytotoxic T-cell PF detected by LDA after lung transplantation, could not be confirmed using flow-cytometric detection of CD8+
T-cell activation and IFN-γ production 
. Therefore, our observation that no loss of donor-specific T-cells occurs after LTx is probably due the higher sensitivity of the CFSE-MLR compared to LDA. Re-stimulation of CFSE-labeled recipient T cells during the last day of culture with CD40-B cells of the same donor allowed quantification of effector function (IFN-γ production) of responding T cells. The data revealed that there is a decrease in the proportions of recipient T-cells producing IFN-γ at 1 year after LTx compared to pre-LTx values. This decrease, however, occurred both upon stimulation with donor-derived and with 3rd
party-derived CD40-B cells, demonstrating that no donor-specific loss of T-cell functionality occurs after LTx. We suppose that the non-specific decrease in the capacity of T cells to produce IFN-γ after LTx is probably due to the continuous treatment with immunosuppressive medication. In summary, our data show that no specific loss of circulating donor-specific T-cell clones occurs during the first year after LTx, that their capacity to expand in response to donor allo-antigens is undisturbed, and that they are still able to mount effector function although at a lower level because of a general impairment of T-cell effector function. These conclusions are consistent with those of Kusaka et al 
, who showed high levels of donor HLA-specific T-cell clonotype mRNAs in PBMC late after renal transplantation. Both studies imply that T cells which recognize donor allo-antigens via the direct pathway remain present in the recipient circulation for at least one year after transplantation. Whether donor-specific hypo-responsiveness might develop later after transplantation will be subject of a future study.
Of the 18 patients studied in the current paper, 4 experienced one or more episodes of acute rejection, occurring between day 10 and day 60 after LTx. The data show some tendency of higher pre-LTx CD3+ and CD8+ donor-specific T-cell PF in the patients which developed acute rejection, but this difference is statistically not significant. The present study was not designed for studying differences between rejectors and non-rejectors, neither powered for that purpose. Associations between donor-specific T-cell PF and acute or chronic rejection will be the subject of a later study with larger numbers of patients.
In conclusion, by using a novel technical approach, we showed for the first time an increase of donor-specific T-cell frequencies shortly after LTx in humans. In addition, we observed that T cells reacting to donor allo-antigens presented via the direct pathway remain present in the recipient circulation for at least 1 year after transplantation.