Alloreactive T cells temporarily increased in the peripheral blood of patients before liver allograft rejection

Abstract

T cells are key mediators of alloresponse during liver transplantation (LTx). However, the dynamics of donor-reactive T-cell clones in peripheral blood during a clinical T-cell–mediated rejection (TCMR) episode remain unknown. Here, we collected serial peripheral blood mononuclear cell samples spanning from pre-LTx to 1 year after LTx and available biopsies during the TCMR episodes from 26 rejecting patients, and serial peripheral blood mononuclear cell samples were collected from 96 nonrejectors. Immunophenotypic and repertoire analyses were integrated on T cells from rejectors, and they were longitudinally compared to nonrejected patients. Donor-reactive T-cell clone was identified and tracked by cross-matching with the mappable donor-reactive T-cell receptor repertoire of each donor-recipient pair in 9 rejectors and 5 nonrejectors. Before transplantation, the naive T-cell percentage and T-cell receptor repertoire diversity of rejectors was comparable to that of healthy control, but it was reduced in nonrejectors. After transplantation, the naïve T-cell percentages decreased, and T-cell receptor repertoires were skewed in rejectors; the phenomenon was not observed in nonrejectors. Alloreactive clones increased in proportion in the peripheral blood of rejectors before TCMR for weeks. The increase was accompanied by the naïve T-cell decline and memory T-cell increase and acquired an activated phenotype. Intragraft alloreactive clone tracking in pre-LTx and post-LTx peripheral blood mononuclear cell samples revealed that the pretransplant naïve T cells were significant contributors to the donor-reactive clones, and they temporarily increased in proportion and subsequently reduced in blood at the beginning of TCMR. Together, our findings offer an insight into the dynamic and origin of alloreactive T cells in clinical LTx TCMR cases and may facilitate disease prediction and management.

INTRODUCTION

Liver transplantation (LTx) is the last clinical resort for patients with end-stage liver disease.^1,2 Nonetheless, T-cell–mediated rejection (TCMR) occurs in up to 10%–30% of recipients of liver transplant on tacrolimus-based immunosuppressive protocols.^3,4 A recent large-scale clinical study reported that acute rejection after LTx significantly increased the risk of graft failure, all-cause mortality, and graft failure–related death.³

Based on the studies of rat liver transplant models, the paradigm of LTx TCMR is that the dendritic cells emerge from the allograft to the recipient lymphoid tissue to activate and expand alloreactive T cells, and these alloreactive T cells are recruited to the graft from the blood circulation and destroy the allograft.^5,6 For humans, the peripheral blood T cells’ proinflammatory cytokine production and their activation status predict the patients at higher risk of TCMR after LTx,^7,8 and repertoire analyses further revealed that the T-cell receptor (TCR) repertoire in blood aligned with the infiltrating TCR repertoire in the context of acute cellular rejection.⁹ However, the donor-reactive T-cell expansion in a TCMR event in patients is obscure.

In this study, we delineated the kinetics of overall T-cell phenotype and TCR repertoire in the peripheral blood of recipients developing TCMR. By tracking the donor-reactive clones in peripheral blood mononuclear cell (PBMC) samples from 5 nonrejectors and 9 rejectors within the period from pre-LTx to 12 months after LTx, we demonstrated the donor-reactive T-cell clones temporarily increased in proportion in peripheral blood before rejection, and the pretransplant naïve T cells were significant contributors to the intrahepatic donor-reactive clones. Our findings may have clinical implications for the clinical management of liver transplant rejection.

METHODS

Approval was obtained from the Ethics Committee of the First Hospital of Jilin University (2019-354). All subjects or legal guardians provided their written, informed consent and assent when appropriate. None of the organs were procured from executed prisoners. Information on the T-cell phenotype monitoring, the sequencing of TCR repertoire, the characterizing of donor-reactive T cells, and statistical analysis is in Supplemental Methods, and Figure S7-S9, http://links.lww.com/LVT/A611.

RESULTS

Peripheral blood naïve T cells declined during TCMR

We enrolled 26 recipients who developed at least 1 episode of TCMR in the first year after surgery; 96 patients who did not undergo TCMR after transplantation were included as nonrejectors (Table 1, Supplemental Table S1, http://links.lww.com/LVT/A611). We first assessed the T-cell phenotype in PBMC samples obtained longitudinally from the 122 HLA-disparate recipients of liver transplant through flow cytometry (Supplemental Table S2, http://links.lww.com/LVT/A611).

TABLE 1.

Patient characteristics

	Patients without TCMR (n = 96)	Patients with TCMR (n = 26)	p
Age recipient (y)
Mean [min–max]	51.09 (27–69)	50.38 (35–68)	0.69
Gender recipient
Male/female	74/22	20/6	0.99
HLA-mismatch
Mean [min–max]	4.25 (2–6)	4.50 (2–6)	0.27
Underlying liver disease
Alcoholic liver disease	16	2
Autoimmune	2	2
Hepatitis B and malignancy	18	6
Hepatitis B–related hepatic cirrhosis	35	8
Hepatitis C–related hepatic cirrhosis	2	1
Primary biliary cirrhosis	3	1
Primary sclerosing cholangitis	3	2
Malignancy	5	1
Other	12	3
Immunosuppressants, n (%)
Basiliximab induction	96 (100)	26 (100)	>0.99
Tac + MMF + Pred	96 (100)	26 (100)	>0.99
Other complication
Biliary complications	10	2	0.47
Infection	16	9	0.90

Abbreviations: LTx, liver transplantation; MMF, mycophenolate mofetil; Pred, prednisolone; Tac, tacrolimus; TCMR, T-cell–mediated rejection.

Before the transplant, we found patients on the list of LTx had significantly lower CD4 and CD8 T-cell numbers in their blood than healthy controls (Figure 1A). After the transplant, the absolute number of both CD4 and CD8 T cells in rejectors and nonrejectors increased significantly (Figure 1A). Compared with nonrejectors whose CD4 and CD8 T cells showed no significant alterations in frequency from pre-LTx to 12 months after LTx (Figure 1B), the frequency of CD8 T cells in rejectors gradually became higher while the frequency of CD4 T cells became lower than nonrejectors in the longitudinal course (Figure 1B), suggesting a CD8-dominant expansion in rejectors after transplantation, which is similar to prior reports.⁹

FIGURE 1.

Longitudinal analysis of T-cell subsets from pre-LTx to 12 months after LTx. (A) Absolute number of lymphocytes. Graphs show the mean ratio (±SEM) of CD4 and CD8 T cells compiled from 122 patients (n = 96 nonrejectors; n = 26 rejectors); gray-shaded rectangles denote 1 SEM around the average ratio of T cells in peripheral blood compiled from healthy controls (n = 8). (B) Frequency of CD4 and CD8 T cells by total T cells. (C) Frequency of T_N cells by CD4 or CD8 T cells. (D) Frequency of T_EM cells by CD4 or CD8 T cells. (E) Absolute number of T_N cells by CD4 or CD8 T cells. (F) Absolute number of T_EM cells by CD4 or CD8 T cells. Significance was assessed using paired t test (paired data) or unpaired t test (unpaired data) with p values <0.05 considered significant. The red asterisk represents the difference between the post-LTx and pre-LTx PBMC samples from rejectors. The blue asterisk represents the difference between the post-LTx and pre-LTx PBMC samples from nonrejectors. The black asterisk represents the difference in PBMC samples between rejectors and nonrejectors. (G) Frequency of T_N cells by CD4 or CD8 T cells of pretransplant PBMC samples (n = 8 healthy controls; n = 96 nonrejectors; n = 22 rejectors; Kruskal-Wallis test) (*p < 0.05; **p < 0.01; ***p < 0.001).

Noticeably, both frequency and absolute number results from pre-LTx PBMC samples showed that the naïve T cell (T_N) level was higher, while the effector memory T cell (T_EM) level was lower in rejecting recipients than in nonrejectors (Figures 1C–F). With respect to subset composition, the pre-LTx CD4 and CD8 T_N cell percentages in rejectors’ T cells were comparable with the healthy donors, while the nonrejectors’ T_N cell percentages were lower than the healthy control (Figure 1G).

After transplantation, the naïve T-cell number showed no significant alterations in absolute number (Figure 1E), but the T_EM number increased in all patients (Figure 1F). The increased T_EM cell counts led to the preoperative difference in the TN and TEM subsets percentage between the 2 groups, which gradually narrowed and exhibited no significant difference at 6 and 12 months after transplantation (Figures 1C, D).

Further, we analyzed the variation in the T-cell subset distribution during TCMR. As TCMR is unpredictable, serial sampling of enrolled recipients was performed during postoperative hospitalization and every follow-up time point. The samples collected at the time point when the liver enzyme level exceeded twice the upper normal range were considered “TCMR” samples, and the samples at 2 consecutive time points before liver enzyme levels exceeded normal levels were considered “pre-TCMR” samples.

Considering the time points for blood draws before and after TCMR was elongated after 2 months, we separately determined the variation in the T-cell subset distribution during TCMR of patients who developed TCMR within 2 months (time interval from pre-TCMR to TCMR was 10.5 days [range: 4–19]), and after 2 months (time interval from pre-TCMR to TCMR was 25.6 days [range: 9–38]).

In all patients, the frequency of CD4 and CD8 T cells and percentage of naïve and memory T cells showed no significant alterations from pre-LTx to pre-TCMR. But after the disease onset, we found the frequency of CD8 T cells increased significantly, and the frequency of CD4 T cells decreased significantly in rejectors (Figure 2A). Furthermore, the decline of naïve T cells and the increase of memory T cells also occurred after TCMR (Figures 2B–E).

FIGURE 2.

T_N in CD4 and CD8 T cells decreased during TCMR. (A) Frequency of CD4 and CD8 T cells by total T cells. (B) Frequency of CD4 and CD8 T_N cells. (C) Frequency of CD4 and CD8 T_EM cells. (D) Absolute number of CD4 and CD8 T_N cells. (E) Absolute number of CD4 and CD8 T_EM cells. (n = 22 rejectors; Friedman test; *p < 0.05; **p < 0.01; ***p < 0.001). Abbreviations: LTx, liver transplantation; TCMR, T-cell–mediated rejection.

Donor-derived T cells were found in recipients’ blood circulation after the operation, but we observed a sustained decline in the donor T-cell chimerism ratio in the first week after LTx, dropping to negligible levels (below 1%) after 2 weeks, suggesting that the T cells in PBMC samples after the first 2 weeks of LTx were mainly of recipient origin (Supplemental Figure S1, http://links.lww.com/LVT/A611).

Collectively, we found that T_N cell percentage in rejectors was comparable to healthy controls, and the proportion of T_N cells reduced during TCMR.

Peripheral blood TCR repertoire of rejectors skewed after LTx

We next analyzed the T-cell dynamic within the longitudinal course at the clone level by bulk TCR sequencing. After excluding samples with insufficient T cells or low yield and integrity of RNA, we successfully constructed TCR repertoires of 22 rejectors and 52 nonrejectors with pre-LTx PBMC samples and constructed serial TCR repertoires of 11 rejectors and 5 nonrejectors with PBMC samples of 1, 6, and 12 months after LTx (Supplemental Table S3, http://links.lww.com/LVT/A611).

We sought to measure the differences among the unique clone distributions. We calculated the inverse Simpson index for the diversity of clones.¹⁰ A significantly higher diversity of the TCR repertoire of the rejectors pre-LTx compared with the nonrejectors was found; however, the TCR repertoires of healthy donors and rejectors showed no statistical differences in diversity (Figure 3A). Next, we calculated the Gini coefficient for inequality among the unique clone distributions.¹¹ Nonrejectors had a higher Gini coefficient than healthy donors and rejectors, and there was also no significant difference in the Gini coefficient between healthy donors and rejectors (Figure 3A).

FIGURE 3.

Overall TCR repertoire of PBMC differs among nonrejectors and rejectors. (A) Inverse Simpson index and Gini coefficient of pretransplant PBMC samples (n = 8 healthy controls; n = 52 nonrejectors; n = 22 rejectors; Kruskal-Wallis test). (B, C) Correlation of inverse Simpson index (B) or Gini coefficient (C) (x-axis) and T_N frequency (y-axis) of pre-LTx PBMC samples from healthy donors and recipients with or without rejection after transplantation (n = 8 healthy controls; n = 52 nonrejectors; n = 22 rejectors). The R value demonstrates the correlation calculated with the Spearman Rank Correlation Coefficient. Significance was assessed using linear regression analysis, and p values <0.05 were considered significant. Each point represents a subject in the study. (D) Longitudinal analysis of inverse Simpson index of PBMC samples after transplantation (n = 5 nonrejectors; n = 11 rejectors; significance was assessed using Friedman test or unpaired t test). (E) Clone sharing at pre-LTx, 1, 6, and 12 months after LTx in nonrejectors (black) and rejectors (red). Morisita’s overlap index of the top 10,000 TCRβ sequences overlaps at each posttransplant time point to the top 10,000 TCRβ sequences pretransplant of allogeneic recipients. (F) The rank of the pretransplant blood-derived top 10 TRBV genes (from top to bottom: top 1st to 10th) tracked after LTx in allogeneic recipients (*p < 0.05; **p < 0.01; ***p < 0.001). Abbreviations: LTx, liver transplantation; PBMC, peripheral blood mononuclear cell; TCR, T-cell receptor.

We further observed the proportions of T_N positively correlated with the TCR repertoire inverse Simpson index and negatively correlated with the Gini coefficient (Figures 3B, C), suggesting the relative abundance of naïve T cells correlated with the TCR repertoire breadth in recipients as expected.

To visualize the clonal diversity for each pre-LTx PBMC sample, we applied “Grouping of Lymphocyte Interactions by Paratope Hotspots 2” (GLIPH2) to identify clusters of TCRs with homologous sequences that potentially recognize the same epitopes.¹² Healthy control had an abundance of TCR sequences that formed more but smaller clusters of clones than nonrejectors. In nonrejectors, the cluster distributions were uneven, and the sizes of the nodes were extremely unequal, which revealed clonal expansion in peripheral blood accompanied by a decrease in diversity before transplantation, whereas the rejectors were located in between (Supplemental Figure S2, http://links.lww.com/LVT/A611).

Intriguingly, there was a progressive loss of TCR repertoire diversity in rejectors after LTx, according to the inverse Simpson index. In contrast, no considerable change in the repertoire breadth of nonrejectors was observed from pre-LTx to after LTx (Figure 3D).

Morisita’s overlap index was employed to quantify the TCR repertoire turnover rate.¹³ For nonrejectors, Morisita’s overlap index of the top 10,000 TCRβ sequences overlap at 1, 6, and 12 months after LTx to the top 10,000 TCRβ sequences pretransplant was stable (almost 1.0), which suggested that the repertoire turnover in most recipients was fairly low, except for P21, who was severely infected with Klebsiella at postoperative day (POD)89 (Figure 3E). Our previous study showed that a recipient who developed graft-versus-host disease at POD60 also had a high rate of turnover (Supplemental Figure S3, http://links.lww.com/LVT/A611).¹⁴ In contrast, Morisita’s overlap index between the postoperative samples and pre-LTx samples in rejectors was low, indicating a drastic overall TCR repertoire turnover during the first year post-LTx in rejectors (Figure 3E). Analysis of the TCRβ variable region (TRBV) gene usage of rejectors also revealed that some of the top 10 V genes pre-LTx were replaced by other genes in rejectors (Figure 3F).

These results indicate that the TCR repertoire in the peripheral blood of rejectors turns over rapidly, accompanied by a reduction of diversity after LTx, which is in agreement with the postoperative tendency of declined T_N and elevated T_EM in rejectors. The phenomenon may reflect the increase of donor-specific clones in peripheral blood.

Donor-reactive clone increased temporarily in peripheral blood before clinical TCMR

To track the kinetics of donor-reactive T cells at the clone level in each subject, we defined the donor-specific clones by mixed lymphocyte reaction (MLR) using donor and recipient pre-LTx PBMC samples, as reported (Figure 4A).^15,16 A schematic showing the tracking of the donor-reactive T cells in samples is shown in Figure 4B. We performed MLR in 23 patients with sufficient donor and recipient pre-LTx PBMC samples as well as serial PBMC samples before and during the TCMR episode; 7 patients were excluded from further analysis due to the poor quality of RNA of the MLR-sorted cells. Two recipients were further excluded from analysis due to insufficient unique TCRβ sequences (<50,000) detected in the MLR-sorted samples. Finally, we tracked the donor-reactive clones in PBMC samples from 5 nonrejectors and 9 rejectors within the period from pre-LTx to 12 months after LTx (Supplemental Table S3, http://links.lww.com/LVT/A611).

FIGURE 4.

Donor-reactive clones increased in proportion in blood before TCMR. (A) Schematic of the mappable TCR repertoire constructed using an MLR with pre-LTx samples. (B) Schematic of the identification and tracking of donor-reactive T cells. (C) Longitudinal data of the transaminase and mappable donor-reactive clones in rejectors. The liver enzymes are plotted on the left y-axis, and the cumulative frequency of mappable HVG clones is plotted on the right y-axis. In the period of rejection, the first point with liver enzyme levels elevated up to more than twice the upper normal range was defined as samples “TCMR.” (D) Longitudinal analysis CD4 and CD8 T_N and T_EM in rejectors. The percentage of recipient-derived T_N (green) and T_EM (orange) (relative to total CD4 or CD8 T cells) in PBMC samples are plotted on the left y-axis, and the cumulative frequency of mappable HVG clones are plotted on the right y-axis. (E) Expression of HLA-DR+ cells by T_EM in rejectors. The frequency of HLA-DR+ cells in T_EM is plotted on the left y-axis, and the cumulative frequency of mappable HVG clones is plotted on the right y-axis. (F) The cumulative frequency of HVG clones in peripheral blood of rejectors at Pre-TCMR and TCMR (n = 9; paired Student t test). Abbreviations: HVG, host versus graft; LTx, liver transplantation; MLR, mixed lymphocyte reaction; PBMC, peripheral blood mononuclear cell; TCMR, T-cell–mediated rejection; TCR, T-cell receptor.

Although the cumulative frequency of posttransplant donor-reactive clones slightly increased in some nonrejectors (P3, P10, and P29), the increase was not remarkable (total frequency of the donor-reactive clones <2%) (Supplemental Figure S4A, http://links.lww.com/LVT/A611, and Supplemental Table S4, http://links.lww.com/LVT/A611). On the contrary, the cumulative frequency of donor-reactive clones in the PBMC samples from rejectors increased remarkably postoperatively (Figure 4C). Although the tacrolimus level of rejectors remained relatively stable before TCMR, HVG (host versus graft) clones increased in proportion significantly (Supplemental Figure S5, http://links.lww.com/LVT/A611). The T-cell subset and donor-reactive T-cell clone dynamics revealed that the cumulative frequency of clone increases coincided with the decrease of T_N and the increase in T_EM preceding the TCMR onset by days or even months in rejectors’ peripheral blood (Figure 4D) and in nonrejectors, reductions in naïve T cells and increase in T_EM were also observed postoperatively; nevertheless, the variation in T-cell subsets did not synchronize with the donor-reactive clone frequency increase in nonrejectors (P3, P10, and P29) (Supplemental Figure S4B, http://links.lww.com/LVT/A611). We investigated whether T cells acquired an activated phenotype during TCMR. Most patients in nonrejectors lacked HLA-DR expression in T_EM cells within the monitoring period (Supplemental Figure S4C, http://links.lww.com/LVT/A611). In contrast, the HLA-DR expression in rejectors was augmented in the peripheral blood T_EM subset, accompanied by an increase in donor-reactive clones before TCMR and the CD8 T_EM to a larger extent than the CD4 counterpart (Figure 4E). However, the donor-reactive clones stopped increasing and significantly decreased at the beginning of the disease onset before the antirejection treatment (Figure 4F), even in P27, who were developed into chronic rejection after the first time TCMR (POD62) and resulted into retransplantation at POD120, suggesting the antidonor clones decreased in blood after TCMR onset even the alloresponse in the liver allograft was not fully suppressed.

Intrahepatic alloreactive T cells originated from preoperative naïve T cells

Considering that the TCR repertoire may differ between the graft and peripheral blood, we first assessed the overall TCR repertoire relevance between the rejecting liver graft and peripheral blood by performing an overlap analysis of TCR repertoires in PBMCs and liver graft biopsies in rejectors. From the rejecting liver tissues sampled from 6 patients with rejection (Supplemental Table S3, http://links.lww.com/LVT/A611), not all intragraft clonotypes could be detected at an equal abundance in their PBMC compartments (Figure 5). Nevertheless, the most dominant clones in the allograft could still be traced in the peripheral blood of these recipients. As expected, the overall TCR repertoire was most relevant to that in peripheral blood taken at biopsy compared with other time points.

FIGURE 5.

Tracking of intrahepatic alloreactive TCRs in peripheral blood. Intrahepatic clones shared between 2 samples with a frequency >10⁻⁵ are shown. Axes denote the frequency of a T-cell clone in the allograft and PBMC samples. The R value demonstrates the correlation between 2 TCR repertoires calculated with Spearman Rank Correlation Coefficient. Significance was assessed using linear regression analysis, and p values <0.05 were considered significant.

To determine the relationship between the circulating and intrahepatic donor-specific T cells, we reidentified the top 10 graft-blood shared clones in a series of corresponding PBMC samples and found that they were rare before LTx (Figure 6A). Frequencies of these clones mostly increased before the rise of liver enzyme levels and decreased at the beginning of TCMR in the peripheral blood, consistent with the PBMC sample results.

FIGURE 6.

Naïve T cells from pre-LTx increased in proportion in allograft. (A) Tracking of top 10 intrahepatic donor-reactive clones shared with blood samples. The time point of TCMR onset is labeled in the red dot line. (B) Stacked bars show the number of distinct donor-specific TCRs that could be recovered only in the naïve or in the memory compartment and those present in both compartments before LTx. (C) The frequency of intrahepatic anti–-third-party TCR repertoire and donor-reactive TCR repertoire. (D) Network of TCR clusters identified by GLIPH2 in liver biopsy and PBMCs at TCMR and annotated by the VDJdb and donor-reactive repertoire. Purple nodes represent donor-reactive sequences identified as being associated with cross-reactive public TCRs, red nodes represent donor-reactive TCRs, blue nodes represent virus-specific clones annotated by VDJdb, and gray-shaded nodes represent TCRβs specific to unknown epitopes. Abbreviations: GLIPH2, Grouping of Lymphocyte Interactions by Paratope Hotspots 2; HVG, host versus graft; LTx, liver transplantation; PBMC, peripheral blood mononuclear cell; TCR, T-cell receptor; TCMR, T-cell–mediated rejection.

To verify if the naïve T cell is the significant contributor to the alloreactive T-cell pool in the rejected allograft, we redefined the donor-specific TCRβ sequences of rejecting allograft in the pre-LTx naïve or memory T-cell repertoire, which were sorted based on the surface expression of CD45RA and CCR7. About 60% of the donor-reactive sequences in the rejected liver could be reidentified in the recipient pre-LTx naïve or memory pool. The majority of donor-reactive TCRβ sequences were recovered in the naïve compartment. Although some alloreactive TCR repertoires were identified as memory subsets in the pre-LTx samples, the number was small (Figure 6B). In the top 10 donor-reactive clones in the biopsy of the 5 recipients, 68% of the clones were recovered in the pre-LTx naïve pool, and 10% were recovered in the pre-LTx memory pool. Our data showed that the naïve TCR repertoire of recipients has a reserve of donor-reactive T cells that could be mobilized during TCMR.

To validate if the expanded intrahepatic T cell is antidonor-specific, the pre-LTx blood samples of the 5 recipients were stimulated in parallel MLR cultures by cells from a third-party donor whose HLA was mismatched with both the donor and recipients. The third-party–reactive T cells showed no posttransplant expansion in the rejected liver and PBMCs in contrast to the antidonor clones (Figure 6C, Supplemental Figure S6, http://links.lww.com/LVT/A611). Considering the recipients had not been sensitized to the third-party allo-antigen, we also calculated the cumulative frequency of cytomegalovirus (CMV) and Epstein-Barr virus (EBV) antigen reactive clones (all 5 recipients were IgG positive and IgM negative) in PBMC samples by querying the sequences in VDJdb.^17,18 No considerable expansion of the CMV or EBV antigen reactive clones in PBMC was found (Supplemental Figure S6, http://links.lww.com/LVT/A611). These results suggested that the expansion of blood and graft is antidonor-specific.

Heterologous cross-reactivity contributes to the pretransplantation allo-specific memory.¹⁹ To investigate whether the intrahepatic donor-reactive TCRβ CDR3 (complementarity determining region 3) sequences have a known antigen specificity for various viruses (CMV, EBV, and influenza virus), thereby to evaluate the possibility of antidonor clone’s preoperative memory T-cell origin, the sequences present in liver biopsy were queried in VDJdb.^17,18 Although the cross-reactive clones were identified, the number was small (Figure 6D). The putative alloantigen-specificity of virus-specific TCR was further determined by clustering analysis of similar TCRβ sequences from biopsies using GLIPH2¹² (Figure 6D). The majority of virus-reactive clones were clustered separately with donor-reactive clones. Together, these results suggested that the pre-existing virus-specific memory clone might contribute less as a donor-reactive one in LTx rejection.

DISCUSSION

Because of the multifactorial and rather complex processes involved, it is still unclear to what extent the pretransplantation naïve and memory T cells contribute to the human LTx rejection. Pre-existing alloreactive memory T cells can be generated through heterologous cross-reactivity and prior alloantigen exposure; they are supposed to be the potential barrier to allograft acceptance because of the ability of direct allograft migration to cause rejection at an early time point without the need for homing to secondary lymphoid tissue²⁰ and the resistance to costimulatory blockade therapy.²¹ However, some reports suggest there is no correlation between cross-reactive memory and rejection in patients.^22,23 Similarly, we found the intragraft virus-specific T cells rarely served as antidonor clones, and the alloreactive T cells in LTx rejectors derived from the pre-existing naïve T cells to a larger extent than pre-existing memory T cells. The large expansion of the naive alloreactive clones to donor-antigen is a possible mechanism that contributes to the clinic LTx TCMR. Basiliximab blocks the IL-2Rα on activated T cells, it also reduces the number of circulating T cells expressing IL-2Rα,therefore it may inhibit the alloresponse mediated by T cells activated shortly after LTx but spare the naïve T cells which need more time to be activated²⁴, thereby basiliximab may have a chance to amplify the role of the pre-existing naïve T cell in TCMR by damping the role of other cells.^25,26 It should also be noted that although CMV and EBV are the most well-known viruses in the alloantigen heterologous cross-reactivity, the pre-existing memory generated by other viruses in patients undergoing LTx should be investigated further. In nonrejectors, the decreased naïve T-cell percentage and TCR repertoire diversity suggested the decrease in the richness of alloreactive TCR repertoire potentially limits the immunity toward donor antigen.

The reduction of TCR repertoire diversity was reported within the first week after LTx.²⁷ We extended the finding by demonstrating that the diversity of TCR repertoire declined during the first year after transplantation in rejecting recipients but not in nonrejectors. Although all patients undergoing LTx exhibited recovery of T-cell count postoperatively, the preferential memory T-cell increase accompanied by a rapid turnover rate in rejectors is not simply a consequence of homeostatic repopulation but reflects the development of donor-reactive T cells.

To understand the donor-reactive T-cell expansion postoperatively, we employed a methodology of identification of donor-reactive clones using MLR established with donor and recipient preoperative PBMC. Using the same method, Savage et al found that the deletion of donor-reactive T-cell clones in blood may be a consequence of LTx instead of a success of immunosuppression withdrawal.²⁸ By more intensively tracking the antidonor clones in the peripheral blood of rejectors, we found the donor-reactive clones were not deleted thoroughly from the blood circulation during the whole period after transplant; they rather increased in blood before the clinical TCMR in rejectors.

Alloreactive clones declined in blood at the beginning of transaminase elevation and before antirejection treatment; we speculated the dynamic was a reflection of the alloreactive clones’ invasion of the liver graft to trigger rejection. By tracking the intrahepatic alloreactive clones in blood, we confirmed the assumption. It should be noted that the donor-reactive clones identified in the study are toward donor MHC; using MLR established with donor and recipient preoperative PBMC would ignore the minor histocompatibility antigens presented in the liver.

Nevertheless, the limitations of this study should be recognized: this is a single-center study with a small cohort of patients, and the RNA-based method prevents donor-reactive clones tracking in some patients with poor RNA quality; further analysis in larger cohorts from other regions will be needed to corroborate our results. Furthermore, we found CD8 T-cell–dominant expansion and activation along with the donor-reactive clone increase in blood, and CD8 T cells are the dominant infiltrators into the rejected liver graft,^29,30 but due to the limited sample volume, we were unable to separately profile the alloreactive CD8 TCR repertoires dynamic.

Our data revealed the origin and expansion dynamics of donor-reactive clones during a clinical TCMR episode. The mechanism of the donor-reactive clones invading the liver graft needs further, more in-depth investigation in animal models by sophisticated cell tracking and bioimaging methods (Table 1).