Overcoming Memory T-Cell Responses for Induction of Delayed Tolerance in Nonhuman Primates


Corresponding author: Tatsuo Kawai, tkawai@partners.org


The presence of alloreactive memory T cells is a major barrier for induction of tolerance in primates. In theory, delaying conditioning for tolerance induction until after organ transplantation could further decrease the efficacy of the regimen, since preexisting alloreactive memory T cells might be stimulated by the transplanted organ. Here, we show that such “delayed tolerance” can be induced in nonhuman primates through the mixed chimerism approach, if specific modifications to overcome/avoid donor-specific memory T-cell responses are provided. These modifications include adequate depletion of CD8+ memory T cells and timing of donor bone marrow administration to minimize levels of proinflammatory cytokines. Using this modified approach, mixed chimerism was induced successfully in 11 of 13 recipients of previously placed renal allografts and long-term survival without immunosuppression could be achieved in at least 6 of these 11 animals.




calcineurin inhibitor


donor bone marrow transplantation


Epstein-Barr Virus


Granzyme B


horse antithymocyte globulin


interfelon γ


kidney transplantation


macrophage inflammatory protein


post-transplant lymphoproliferative disease


reverse transcription polymerase chain reaction


simultaneous kidney and bone marrow transplantation


central memory T cell


effector memory T cell


thymic irradiation


memory T cells


Memory T cells (Tmem), elicited against environmental pathogens, vaccines, paternal antigens during pregnancy or transfusions, frequently cross-react with unrelated pathogens or alloantigens. This process, known as heterologous immunity (1–3), constitutes a potent barrier to tolerance induction or maintenance by the mixed chimerism approach (4) or by costimulatory blockade (5). Homeostatic proliferation of Tmem observed following T-cell depletion also generates alloreactive Tmem and may thereby create resistance to tolerance induction by costimulatory blockade (6). Most clinical tolerance trial regimens involve profound T-cell deletion (7–9) and therefore could be negatively affected by homeostatic proliferation (6). Thus, Tmem responses could constitute a major barrier to tolerance induction (4,5,10,11).

We previously developed a nonmyeloablative preparative regimen that permits the induction of mixed chimerism and renal allograft tolerance following peritransplant conditioning and simultaneous kidney and donor bone marrow transplantation (SKBMT) in MHC mismatched cynomolgus monkeys (12–14). This approach was successfully extended to HLA matched or mismatched human kidney transplant recipients (15,16). Conditioning for this protocol requires recipient treatment beginning 6 days prior to the planned SKBMT, thereby limiting applicability to living donor transplant recipients. A major goal of the study reported here has been to establish a comparably effective “delayed tolerance” regimen, in which recipients initially undergo kidney transplantation (KTx) with conventional immunosuppression and subsequently receive nonmyeloablative conditioning and donor bone marrow transplantation (DBMT) from the original kidney donor. However, “delayed tolerance” has the theoretical disadvantage that donor-specific Tmem might be elicited against the allograft even during administration of potent suppressive agents. Here, we show that donor-specific allo-reactive Tmem responses constitute a potent obstacle to establishment of allograft tolerance induction in this stringent model. Nevertheless, induction of renal allograft tolerance can be achieved by specific modifications to overcome these Tmem responses.



Fifty male cynomolgus monkeys that weighed 3–7 kg were used (Charles River Primates, Wilmington, MA, USA). All surgical procedures and postoperative care of animals were performed in accordance with National Institute of Health guidelines for the care and use of primates and were approved by the Massachusetts General Hospital Subcommittee on Animal Research.

Cynomolgus MHC genotyping

First, genomic DNA was prepared from PBMC and splenocytes. Panels of seventeen microsatellite loci spanning ∼5 Mb of the MHC region were amplified from the genomic DNA with fluorescent-labeled PCR primers and fragment size analysis was determined. The microsatellite haplotypes for each animal were converted to predicted MHC genotypes based on our previous cloning and sequencing work with cynomolgus monkeys (17,18).

Conditioning regimens (Figure 1)

All recipients initially underwent KTx alone with a conventional triple drug immunosuppressive regimen consisting of tacrolimus (Astellas Pharma Inc. Osaka, Japan), mycophenolate mofetil (Roche Inc., Nutley, NJ, USA) and prednisone. The recipients then underwent conditioning and DBMT 1–4 months later. The standard conditioning regimen consisted of low-dose total body irradiation (TBI, 1.5 GyX2) on days −6 and −5, thymic irradiation (TI, 7 Gy) on day −1, equine ATG (ATGAM, Pharmacia and Upjohn, Kalamazoo, MI, 50 mg/kg/day on days −2, −1 and 0) and anti-CD154 monoclonal antibody (anti-CD40L, American Type Culture Collection catalog number 5c8.33, 20 mg/kg on days 0 and +2, and 10 mg/kg on day 5, 7, 9 and 12).

Group A: The recipients received the standard conditioning regimen 4 months after KTx. Group B: A course of humanized anti-CD8 mAb administration (cM-T807 provided by Centocor Inc., Horsham, PA, USA) was begin at 5 mg/kg on Days 0 and 2 after BMT in addition to the standard regimen. Group C: Recipients were treated exactly as in group B, but the conditioning and BMT were introduced at 1 month after KTx. All recipients (except for one group B recipient who later died due to right atrial thrombosis) were pretreated with Ketorolac (1 mg/kg intravenous injections on days −1 and 0) to prevent anti-CD154 mAb induced thrombosis, as previously described (19). The preliminary data in groups A and B were partially reported previously (20)

Renal transplantation

The detail of KTx procedure was reported previously (21). The recipients also underwent unilateral native nephrectomy and ligation of the contralateral ureter on day 0. The remaining native (hydronephrotic) kidney was removed between 60 and 80 days after transplantation.

Bone marrow transplantation

Bone marrow was harvested from donor iliac bones by multiple percutaneous aspirations. If the donor animal was sacrificed, bone marrow cells were also harvested from the vertebral bones after the euthanasia and kept frozen until BMT. These cells (1.0–3.0 × 108 mononuclear cells/kg) were infused intravenously.

Flow cytometric analyses, detection of chimerism and cell sorting

PBMC, peripheral lymph nodes, spleen and bone marrow cells were labeled with a combination of the following mAbs: CD3 PerCP (SP 34–2), CD4 PerCP (L-200), CD8 PerCP (SK1), CD8 APC (SK1), CD95 FITC (DX2), CD95 APC (DX2), and CD28PE (CD28.2) (BD Pharmingen, San Jose, CA, USA). Lymphocytes from the animals treated with anti-CD8 mAb were stained with anti-CD8-PE (DK25, Dako, Inc., Carpenteria, CA, USA). For chimerism analysis, we used a donor-specific mAb chosen from a panel of mouse antihuman HLA class I mAbs (H38, One Lambda, Inc., Canoga Park, CA, USA) that cross-react with cynomolgus monkeys. The recipient and donor pairs were chosen based on the reactivity against the mAb. The fluorescence of the stained samples was analyzed using FACS Calibur and FACS Scan flow cytometers and Cell Quest Software (BD), or FlowJo software. For assessing memory cell function, fresh PBMCs were gated on lymphocytes and sorted into CD95-CD28 + naive and CD95+CD28 low/high memory populations using a FACS Vantage cell sorter (BD Immunocytometry System). The purity of sorted cells was consistently > 95%.

Measurement of alloresponses by ELISPOT

ELISPOT plates (Millipore, Bedford, MA, USA) were precoated with 5 μg/mL of capture antibodies against IFNγ, and IL-2 (Mabtech, Sweden) in PBS and stored overnight at 4°C. The responding cells separated from fresh PBMC were cocultured with an equal number of irradiated donor PBMCs as stimulating cells (1.5 × 105 cells/well), or unstimulated in medium alone, or with PHA at 1 μg/mL (Sigma). After 20 and 44 h incubation at 37°C for IL-2 and IFNγ respectively, the plates were washed and biotinylated detectionantibodies (Mabtech, Sweden) were added (4°C OVN). After five washes with PBS, streptavidin–horseradish–peroxidase conjugate in PBS BSA 0.5% (Dako, Glostrup, Denmark) was added for 2 h at room temperature, followed by five washes. Thereafter, 50 μL/well of 3,3,5,5-tetramethylbenzidine (TMB) liquid substrate (Sigma-Aldrich) was added and incubated for 30 min in the dark. The resulting spots were counted with an ELISPOT image analyzer (CTL Inc., Cleveland, OH, USA).

Alloantibody measurement

Antidonor alloantibody was detected by flow cytometric analysis. Donor PBMCs were first incubated with recipient sera for 30 min at 4°C. After washing with FACS medium, FITC conjugated mouse antihuman IgG mAb was added and incubated for 30 min at 4°C, and then washed twice. PBMCs were further incubated with PE conjugated anti-CD20 mAb (Becton Dickinson, Mountain View, CA, USA) for 30 min at 4°C. After washing, PBMCs were fixed with 2% paraformaldehyde. Cells were then acquired and analyzed with FACScan (Becton Dickinson). A positive reaction of the T and B cells was defined as a shift greater than 10 channels in mean fluorescence (MCF) intensity of donor lymphocytes when using test sera compared with pretransplant serum control.

Quantitative real-time PCR analysis

Peripheral blood was collected in PaxGene RNA collection tubes (BD Diagnostics, Valencia, CA, USA), allowed to sit at room temperature for 6–24 h, and stored at −700C until RNA extraction. Total RNA was extracted from peripheral blood cells by a PaxGene blood miRNA kit (PreAnalytiX, Qiagen, Hilden, Germany) and the RNA concentration was measured using a Nanodrop ND 1000 spectrophotometer (Wilmington, DE, USA). One microgram of total RNA was converted into cDNA using a Taqman Reverse Transcription kit (Applied Biosystems Inc., Foster City, CA, USA). Qt-RT-PCR analysis was performed by a two-step process, a 10-cycle preamplification step (AmpliTaq® DNA Polymerase Kit; Applied Biosystems Inc.) followed by measurement of mRNA quantity (StepOnePlus Real-Time PCR System, Applied Biosystems Inc.). Primer-probe (P&P) sets were custom designed for the measurement of mRNA levels of cytokines (Supporting Table S1). QuantiFast Probe PCR Master Mix was purchased from Qiagen (Hilden, Germany) and qt-RT-PCR was performed. Amplification was carried out in a total volume of 20 μL each in duplicate wells for all samples, with an initial incubation for 3 min at 95°C followed by 40 cycles of 3 s at 95°C, 30 s at 60°C. Transcript copy numbers were calculated by a standard curve method (22). Target gene expression was normalized by 18S rRNA (house keeping gene (HKG)) expression. For the samples with <25 copies, a value of 12.5 copies was assigned, which is half the minimum copies used on the standard curve. And for the samples with no detectable copies of transcripts (zero), half the minimum observed 18S-rRNA normalized copies for the specific gene were used.

Statistical analysis

Alloreactive memory responses and mRNA for cytokine profiles were analyzed by a nonparametric analysis Mann–Whitney–Wilcoxon (MWW) signed-rank test.

A renal allograft survival curve was created by the Kaplan and Meier method and results were compared using the log rank test (Prism version 4, Graphpad Software Inc., San Diego, CA, USA).


Prolonged deletion of CD8+ Tmem is required for induction of chimerism in “delayed tolerance”

As reported in a preliminary study (20), the current study with larger number of recipients concluded that prolonged deletion of CD8+ effector memory T cells (CD8 TEM, CD3+CD8+CD95+CD28−) is required for successful induction of chimerism in the delayed protocol. In group A (Table 1), five recipients initially underwent KTx alone with conventional maintenance immunosuppression including a calcineurine inhibitor, mycophenolate mofetil and prednisone. Four months after KTx, these recipients were conditioned with a nonmyeloablative regimen and received DBMT from the kidney donor (Figure 1). In group A, marked homeostatic expansion of CD8+ TEM was observed by day 15–20 (Figure 2) and none of the recipients developed multilineage chimerism (Figure 3) or renal allograft tolerance. One of the five recipients developed limited hematopoietic chimerism only in the myeloid lineage and the transplanted kidney survived long term (703 days) but with chronic rejection detected by day 100 after DBMT (Table 1).

Table 1.  Conditioning regimens and results of chimerism/allograft survival
 NTiming of BMT relative to KTxaCD8 mAbMultilineage chimerism (MC1) Kidney survival after DBMT
  1. 1MC, multilineage (lymphoid and myeloid lineages) chimerism detectable by flow cytometry, 2chronic rejection, 3acute rejection, 4transplant glomerulopathy, 5PTLD, 6bowel obstruction, 7right atrial thrombosis, 8BK infection, 9CMV infection, 10sepsis.

A5Day 119 (4 months)0/5MC−7032,3, 643, 375, 353, 306
B13 Day 119 (4 months)+11/13MC+18254, 852, >849, >744, >555, >520, 753, 755, 627, 325, 148
     MC−1073, 673
C7Day 28 (1 month)+7/7MC+1803, 1253, 713, 693, 269, 209, 1910
Figure 1.

Conventional Immunosuppression after KTx and nonmyeloablative conditioning regimen for DBMT. All recipients initially underwent KTx alone with a conventional triple drug immunosuppressive regimen consisting of tacrolimus, mycophenolate mofetil and prednisone. The recipients then underwent conditioning and DBMT 1 (group C) to 4 months later (groups A and B). The standard conditioning regimen consisted of low-dose total body irradiation (1.5 GyX2) on days −6 and −5, thymic irradiation (TI, 7 Gy) on day −1, equine ATG (hATG) and anti-CD154 mAb. In groups B and C, additional anti-CD8 mAb was also administered. After DBMT, the recipient was treated with cyclosporine for 28 days, after which no further immunosuppression was administered. KTx = kidney transplantation; TBI = total body irradiation; TI = thymic irradiation; DBMT = donor bone marrow transplantation; CNI = calcineurin inhibitor; hATG = equine antithymocyte globulin.

Figure 2.

Lymphocyte subsets after DBMT. There was no significant difference observed in CD4 T-cell subsets among groups A, B and C. In group A, although naive CD8+ T cells were easily deleted, both CD8+ TEM and CD8+ TCM started to recover rapidly by day 5. In groups B and C, by adding humanized anti-CD8 mAb (cMT807), prolonged deletion of both CD8+ TEM and CD8+ TCM were achieved for 4 weeks. There was no difference observed in T-cell subsets between groups B and C. There were significant differences in CD8+TEM and CD8+TCM between group A versus groups B or C (*p < 0.05, **p < 0.01).

Figure 3.

Mixed chimerism after DBMT in groups A, B and C. Chimerism (%) in the myeloid (A) and the lymphoid (B) lineages: in group A (n = 5), 4/5 failed to develop mixed chimerism. One monkey in group A developed limited chimerism only in the myeloid lineage. By adding humanized anti-CD8 mAb (cMT807) to the conditioning regimen in group B (n = 13), 11/13 recipients successfully developed multilineage mixed chimerism. In group C (n = 7), DBMT was performed at 1 month after KTx with same conditioning used in group B. Excellent chimerism was also induced in both myeloid and lymphoid lineages in group C.

In group B, two doses of humanized anti-CD8 monoclonal antibody (cM-T807) were added to the conditioning regimen (Figure 1) in 13 recipients. The addition of cM-T807 to the conditioning regimen prevented the rapid homeostatic expansion of CD8+ TEM until day 30 (Figure 2) and 11/13 recipients successfully developed multilineage chimerism (Figure 3). Although chimerism eventually declined and became undetectable by day 60, 6 of the 11 recipients with this transient multilineage chimerism achieved long-term renal allograft survival without maintenance immunosuppression (Table 1, Figures 4 and 5). In the remaining five recipients who had chimerism, only one developed rejection. Other four recipients died due to EBV-related PTLD (two recipients), CMV infection and right atrial thrombosis (due to no Ketorolac prophylaxis at DBMT). The last two recipients in this group failed to develop chimerism and lost their kidney allografts due to acute rejection shortly after discontinuation of immunosuppression (Table 1 and Figure 4).

Figure 4.

Serum Creatinine after donor bone marrow transplantation (DBMT). In group A, none of the recipients achieved renal allograft tolerance. One of the five recipients developed limited hematopoietic chimerism and the transplanted kidney survived long term (703 days) but with chronic rejection detected by day 100 after DBMT. In group B, 11 recipients developed chimerism and six of them achieved long-term renal allograft survival without maintenance immunosuppression. In group C, none of recipients achieved long-term allograft survival.

Figure 5.

Histopathology of kidney allografts in long-term survivors (group B). Renal biopsies from recipients with functioning allografts 9–24 months post-transplant. (A) M1902 day 1825 autopsy showed transplant glomerulopathy and focal segmental glomerulosclerosis (×40). Although alloantibody and C4d deposition were negative, this monkey started to develop transplant glomerulopathy after day 1100. (B) M4403 day 852 autopsy showed no rejection (×20). Euthanized due to nonimmunological reason with normal kidney function. (C) M8907 at 2 years post-transplant shows normal histopathology. The capsule is at the upper right. (D) M3208 at 1 year post-transplant shows a moderate diffuse mononuclear infiltrate with mild glomerulitis and segmental GBM duplication (transplant glomerulopathy) not well seen at this magnification. (E) M2108 at 9 months shows a focal mononuclear aggregrate. The glomeruli are normal and there is no evidence of rejection. (F) M6007 at 2 years shows hypertrophied glomeruli with GBM duplication and a minimal interstitial infiltrate. All PAS stains; original magnifications ×100. C4d stains were negative on all the biopsies (not shown).

Earlier timing of delayed DBMT failed to induce allograft tolerance

In group C, seven recipients were treated identically to those in group B, except that nonmyeloablative conditioning and BMT were performed at 1 month rather than 4 months post-KTx. Lymphocyte subsets in group C recipients were not significantly different from those in group B (Figure 2) and all six recipients in group C developed excellent chimerism, comparable to that observed in group B recipients (Figure 3). Nevertheless, four monkeys developed acute cellular rejection without alloantibody and none of the group C recipients achieved long-term allograft survival (Table 1, Figure 4). Significantly better renal allograft in group B (Figure 6A) could not be attributed to a better donor/recipient histocompatibility, since there was no correlation between donor/recipient MHC matching and the renal allograft survival (Figure 6B).

Figure 6.

Renal allograft survival and MHC matching. To evaluate immunological outcome alone, death due to virus infection or PTLD were censored. (A) Renal allograft survival in groups A, B and C: Kaplan–Meier analysis showed significantly superior renal allograft survival in group B versus group A (p = 0.0199) and group C (p = 0.0338). (B) MHC mismatch and renal allograft survival: Genotypes of MHC class I and class II of all recipients and donors were identified by the microsatellite assay (17,18). Recipients in groups B and C were grouped by the number of total MHC mismatches (A, B, DR and DP) with donor MHC. No correlation was found between the number of MHC mismatches and the renal allograft survival. mis: number of MHC mismatch. Group B 4 mis (n = 4), group B 5 mis (n = 3), group B 6–8 mis (n = 4), group C 4 mis (n = 2) and group C 6–8 mis (n = 2).

Time-dependent reduction of allo-reactive Tmem responses after KTx

The frequencies of Tmem were measured by IFNγ and IL-2 spots with ELISPOT, in which FACS-sorted Tmem were cocultured with irradiated donor and third party cells for 44 and 20 h, respectively (23). Recipients showed various magnitude (from 50 spots to >2800 spots/million Tmem) of antidonor Tmem responses before KTx. After KTx, but before BMT, during the period of maintenance immunosuppression, we observed a time-dependent reduction of donor-specific Tmem responses. Figure 7A details sequential IFNγ Tmem responses (either CD95+ sorted cells or bulk peripheral blood mononuclear cells) after KTx from a representative monkey. High levels of antidonor Tmem responses persisted until 2 months post-KTx, after which a significant decrease was observed by 4 months. Therefore, antidonor Tmem responses were still significantly high at 1 month when group C recipients underwent conditioning and DBMT. In contrast, group B recipients underwent DBMT at 4 months, when antidonor Tmem responses measured by IFNγ secretion became significantly lower than in group C (Figure 7B). Although statistically not significant, a similar trend was observed in IL-2 Tmem responses between groups B and C (Figure 7B).

Figure 7.

(A) Antidonor Tmem responses after KTx. IFNγ spots in ELISPOT by either CD95+ sorted cells or bulk peripheral blood cells were measured sequentially after KTx in a representative recipient. Significantly high antidonor Tmem responses were detectable until 2 months after KTx. However they waned thereafter and became significantly lower by 4 months. (B) Antidonor Tmem responses before KTx and DBMT. Frequencies of Tmem (CD95+ cells) secreting IFNγ (left panel) and IL-2 (right panel) were measured by ELISPOT before KTx and immediately before DBMT. Antidonor Tmem responses were not different between groups B and C before KTx (gray column). However, antidonor Tmem responses waned after KTx in time-dependent fashion and decreased to almost undetectable levels by 4 months. Therefore, Tmem responses before DBMT was significantly lower (IFNγ p = 0.028) in group B (DBMT at 4 months) than in group C (DBMT at 1 month). IL-2 Tmem responses were also not detectable in group B immediately before DBMT. (C) Anti-3rd party Tmem responses before KTx and DBMT. In contrast to antidonor Tmem responses, anti-3rd party Tmem responses were relatively preserved after KTx (black column) and significant numbers of IFNγ and IL-2 spots were detectable even at 4 months after KTx (group B).

Interestingly, this reduction in Tmem responses was donor specific as Tmem responses against the third party were relatively preserved even in group B (at 4 months) (Figure 7C).

Tmem response after DBMT; tolerant recipients versus rejectors

Serial monitoring of Tmem responses was performed after DBMT in four group B recipients who survived long term without immunosuppression. Tmem responses in all long-term survivors remained unresponsive against the donor, while vigorous responses were observed against the third-party cells (Figure 8A). In contrast, three rejectors in group C showed vigorous antidonor Tmem responses after DBMT (Figure 8B).

Figure 8.

Tmem responses after DBMT. Serial monitoring of Tmem responses were performed after DBMT in four group B recipients who survived long-term without immunosuppression. Tmem responses in all long-term survivors remained unresponsive against the donor, while vigorous responses were observed against the third party cells (A). In contrast, three rejectors in group C showed vigorous antidonor Tmem responses after DBMT (B).

High inflammatory responses at 1 month after KTx were associated with renal allograft rejection

T cells take cues from the inflammatory milieu (24,25) and the texture of the inflammatory environment can impact the molecular phenotype and function of alloreactive T cells, which could be a cause of residual Tmem responses and reactivation and/or generation of donor-specific Tmem at 1 month after KTx (group C). Thus, we next evaluated various inflammatory cytokine responses in the peripheral blood before and after DBMT by measuring their mRNA levels (by quantitative real-time PCR (qt-RT-PCR)) or serum levels by the Luminex assay. The list of cytokines or costimulatory pathways measured by qt-RT-PCR included Granzyme B (GZMB), IL-17F, IL-12, IL8, TNFα, RANTES, MIP1α, MIP1β, IL-1β, IL6, CD40, CD80 and CD86. The list of serum cytokines measured by Luminex included IFNγ, IL10, IL12, IL15, IL17F, IL-1β, IL-1Rα, IL-2, IL-4, IL-5, IL-6, MIP1α and TNFα. Among various inflammatory cytokines or costimulatory pathways, significantly more robust mRNA expression of IL-6, IL-17F, IL-1β and GZMB was detected immediately before DBMT conditioning in group C, compared to that in group B (Figure 9A, p = 0.008, 0.0167, 0.0167, 0.009, respectively). Among various serum cytokines measured by Luminex, only serum IL-6 level was statistically higher in group C immediately before DBMT conditioning (p = 0.004). The serum levels of IL-17F were also higher in group C but did not reach statistical significance (Figure 9B). These observations suggest that higher inflammatory environment might have led to higher alloreactive Tmem responses, which resulted in failure of tolerance induction.

Figure 9.

Inflammatory responses after KTx. (A) Messenger RNA (RT-PCR): Peripheral blood samples were obtained from 10 recipients in group B and 6 recipients in group C before DBMT. Total RNA was extracted from recipient whole peripheral blood collected in PaxGene tubes and qt-RT-PCR was performed using cytokine specific primers and probes. 18S was used as a housekeeping gene and copies of cytokine transcripts were normalized by copies of 18S. Results are expressed as ratio of cytokine copy number to 18S. Among various cytokines tested, mRNA for inflammatory cytokines IL-6, IL-17F, IL-1β and Granzyme B (GZMB) were significantly higher before BMT in group C compared to those in group B (p = 0.008, 0.0167, 0.0167 and 0.009, respectively). The value in Y axis was expressed as ratio of copy number to 18s. (B) Serum cytokine levels (Luminex): Among various serum cytokines measured by Luminex, only serum IL-6 levels were statistically higher in group C immediately before DBMT conditioning (p = 0.004). The serum levels of IL-17 were also higher in group C but did not reach statistical significance.


Naive cynomolgus monkeys show various levels of heterologous Tmem responses that are cross-reactive with allogeneic MHC molecules. Our previous study showed that the majority of allospecific Tmem secreting IL-2 are CD4+ central Tmem, while most of the T cells producing IFNγ correspond to CD8+ effector Tmem (23). Significant numbers of Tmem producing either IFNγ or IL-2 could be detected as early as 8 h after allogeneic stimulation. The kinetics are quite different with naive T cells for which the direct alloresponse is detectable only 24 h after alloantigen stimulation and reached a maximum level at 48–72 h (23). Disparity in MHC gene matching is generally associated with a higher frequency of donor-reactive memory T cells, which can potentially affect the long-term transplant outcome (23).

In “delayed tolerance”, we predicted that these heterologous memory responses would be further stimulated by the kidney allograft, and might make tolerance induction more difficult. However, somewhat to our surprise, during the period of maintenance immunosuppression with conventional triple immunosuppression following KTx, donor-specific alloreactive Tmem responses decreased in a time-dependent manner. This was not simply due to global effects of immunosuppression, since anti-3rd party responses were more preserved than antidonor responses. Such donor-specific Tmem hyporesponsiveness has also been reported in clinical kidney transplantation (26,27) and is speculated to result from the interaction between recipient lymphocytes and tolerogenic graft parenchymal cells (28). In those studies, T-cell hyporesponsiveness was observed to develop predominantly in the Tmem subset (26), possibly due to the fact that Tmem is the only subset capable of migrating across vascular endothelium to reach the allograft parenchyma (29). Furthermore, there is evidence suggesting that antigen exposure beyond the initial expansion phase can lead to Tmem exhaustion under certain circumstances (30). In chronic viral infection, high viral load, sustained high proinflammatory cytokine production and/or lymphocyte infection have been postulated to induce Tmem exhaustion (31). Recent report also demonstrated that chronic antigen exposure alone is sufficient to induce exhaustion in both CD4 and CD8 Tmem (32,33). Since donor-specific reduction of Tmem responses appeared time dependent, the residual Tmem function was still detectable at 1 month after KTx (group C), which may have resulted in the failure of tolerance induction.

However, even with nearly undetectable level of alloreactive Tmem before DBMT, prolonged depletion of CD8+ Tmem was still required for successful chimerism induction in the delayed tolerance protocol (20). Without anti-CD8 mAb therapy, homeostatic expansion of CD8+ Tmem ensued by day 15 and all recipients failed to develop mixed chimerism. In contrast, homeostatic expansion of CD8+ Tmem was delayed until day 30 in groups B and C and recipients in these groups developed multilineage chimerism. Our observations suggest that CD8+ Tmem responses could be reactivated by the donor bone marrow cell infusion, even if Tmem responses declined to the nearly undetectable level by ELISPOT. In considering clinical application, since CD8+ Tmem is critical for antiviral immunity (34), the recipients were closely monitored for evidence of new viral infections. In fact, some recipients did develop lethal virus infection or EBV-related lymphoma. Comparing the incidence of viral infection in the previous study without anti-CD8 mAb (14), viral infection (especially CMV) was clearly increased in this series of experiments. However, anti-CD154 mAb has also been changed to humanized one (previously mouse mAb) and it is not certain that higher incidence of infection was caused by addition of anti-CD8 mAb alone. To decrease the incidence of infection, the regimen will probably require further revision with more specific suppression of antidonor Tmem that does not abrogate antiviral immunity.

Attempts to define possible explanations for residual Tmem responses, in the recipients that failed to achieve long-term renal allograft survival following delayed treatment at 1 month, revealed significantly higher mRNAs of inflammatory cytokines such as IL-1β, GZMB, IL-6, IL-17F before DBMT in group C than in group B. Significantly higher serum IL-6 levels were also detected by Luminex. Perhaps elevated levels of these inflammatory cytokines at 1 month after KTx compared to 4 months after KTx reflects the persistence of ischemic/reperfusion injury of the kidney allograft. Heightened innate immunity is activated through a toll-like receptor, which leads to inflammatory cytokine production, upregulation of costimulatory molecules, maturation of dendritic cells and activation of the adaptive immune system (35–37). Moreover expression of IL-6 prevents recruitment of naive T cells to the Foxp3+ immunoregulatory phenotype (24) and favors commitment into the Th17 phenotype. Furthermore, the host environment is critical in regulating the alloresponses and local tissue inflammation is associated with the recruitment of alloreactive T cells in the mixed chimera model (38). Thus, when DBMT was performed at 1 month after KTx, these residual Tmem responses were easily activated by the allogeneic bone marrow cells and eventually rejected the kidney allograft after discontinuation of immunosuppression. Such inflammatory responses and antidonor memory responses appeared to have subsided to almost undetectable levels by 4 months after KTx. Thus, timing of BMT appeared to be critical for successful induction of “delayed tolerance” in our approach. Studies evaluating blocking of these inflammatory cytokine responses by adding anti-inflammatory agents, such as α1-antitrypsin or anti-IL-6R mAb, to the conditioning regimen are now underway.

In conclusion, “delayed tolerance” can be achieved through the mixed chimerism approach but with specific modifications to overcome/avoid alloreactive Tmem responses. High levels of pretransplant heterologous Tmem responses subsided in a time-dependent manner after KTx with appropriate immunosuppression. However, for successful delayed induction of tolerance, the nonmyeloablative conditioning with additional anti-CD8 mAb should be initiated at the optimal timing, when antidonor memory responses subside with less inflammatory milieu being established.


This study was supported in part by 5U19DK080652-02 NHL-BI, POI-HL18646, NIH-NIAID, ROIA137692 and NIH/NIAID 5R01 AI50987-03 and NIH R00000000008108.

We thank Ms. Patricia Della Pelle, Susan Shea and Kento Kawai for technical assistance and Drs. Michael Duggan and Elisabeth Moeller for anesthesia and postoperative care.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation. This manuscript was also not prepared or funded by any commercial organization.