Partial T cell depletion is used in solid organ transplantation as a valuable strategy of peritransplant induction immunosuppression. Using a murine cardiac allograft model, we recently demonstrated that this led to lymphopenia-induced (homeostatic) proliferation among the residual nondepleted lymphocytes. Rather than promoting tolerance, peritransplant T cell-depleting Abs actually resulted in resistance to tolerance induction by costimulatory blockade. In this study we show that memory T cells predominate shortly after subtotal lymphodepletion due to two distinct mechanisms: relative resistance to depletion and enhanced homeostatic proliferation. In contrast, regulatory cells (CD4+CD25+Foxp3+) are depleted as efficiently as nonregulatory cells and exhibit reduced homeostatic expansion compared with memory cells. The resistance to tolerance induction seen with subtotal T cell depletion can be overcome in two different ways: first, by the adoptive transfer of additional unprimed regulatory cells at the time of transplant, and second, by the adjunctive use of nondepleting anti-CD4 and anti-CD8 mAbs, which effectively block homeostatic expansion. We conclude that the resistance to tolerance induction seen after subtotal lymphocyte depletion can be attributed to alterations in the balance of naive, memory, and regulatory T cells. These data have clinically relevant implications related to the development of novel strategies to overcome resistance to tolerance.
Donor-specific tolerance to transplanted tissues remains an elusive goal in clinical transplantation. At the same time, a number of distinct protocols have been reported in rodent models that induce robust tolerance. These include mixed chimerism (1) and costimulatory blockade (2, 3, 4). Lymphocyte depletion with concomitant use of donor bone marrow has also been used successfully, without an obvious requirement for sustained hemopoietic chimerism (5). This finding has generated support for the use and study of lymphodepleting agents as induction therapy in clinical settings, even without the use of donor marrow.
Several years ago it was recognized that T cells undergo extensive proliferation under conditions of lymphopenia (6). This phenomenon, termed homeostatic proliferation or lymphopenia-induced proliferation, is driven by self-MHC/peptide complexes and γ-chain cytokines such as IL-7 and IL-15 (6, 7, 8, 9). Recently, we used a murine cardiac allograft model to study whether homeostatic proliferation might occur in the context of clinically relevant lymphodepletion and what impact this might have on susceptibility to tolerance (10). We found that residual lymphocytes undergo homeostatic expansion in partially depleted hosts. Moreover, rather than promote tolerance, partial T cell depletion actually resulted in tolerance resistance, which we showed was dominantly transferable to naive animals.
A growing body of literature shows that lymphocytes emerging in the aftermath of lymphopenia-induced homeostatic proliferation acquire characteristics of effector/memory cells (11, 12, 13, 14, 15). Hence, one explanation for the rejection that we observed in our study is that homeostatic proliferation led to a skewing of the lymphocyte pool toward a higher percentage of circulating effector or memory cells, which are intrinsically more resistant to tolerance induction (16, 17). Alternatively, Abs have been suggested to deplete naive and/or regulatory T cells (Tregs)2 more efficiently than memory cells (18, 19). It is also possible that residual memory cells, which are not eliminated, may undergo more rapid expansion compared with residual naive cells in the aftermath of lymphodepletion. Either of these two possibilities would lead to an immune repertoire skewed toward an effector/memory phenotype. These hypotheses are not mutually exclusive, and all could contribute to tolerance resistance.
In this study we examine the T cells that emerge in the aftermath of depletion and show that they are skewed toward the memory phenotype because of both resistance to depletion as well as enhanced homeostatic proliferation compared with naive T cells or CD4+CD25+Forkhead/winged helix transcription factor (Foxp3) Tregs. We show that despite this increase in memory cells after partial lymphodepletion, tolerance can be restored though the use of exogenous Tregs or the administration of nondepleting anti-T cell Abs, which we find act to block homeostatic expansion in the setting of lymphopenia. Thus, the resistance to tolerance induction seen after subtotal lymphocyte depletion can be attributed to alterations in the balance of naive, memory, and regulatory cells. Our data also suggest that clinical protocols using lymphocyte depletion may also be enhanced by the use of exogenous Tregs or maneuvers to promote their recovery and/or agents that block homeostatic expansion, such as nondepleting anti-T cell Abs.
Materials and Methods
Mice and Ab treatment
C57BL/6 (Thy1.2+), C57BL/6 scid, C57BL/6 (Thy1.1+), C57BL/6 (CD45.1+), and BALB/c mice (The Jackson Laboratory) were maintained in pathogen-free conditions at the University of Pennsylvania animal facility. Subtotal T cell depletion was achieved using the anti-CD4 Ab GK1.5 and the anti-CD8 Ab 2.43 (both obtained from Bioexpress; 100 μg/dose) via i.p. injection on days 0 and 3. This protocol has previously been shown to cause ∼85% CD4 and CD8 depletion (10). Mice were killed, and lymphocytes were isolated at various time points beginning 5 days after the first dose of depleting mAb.
Six- to 8-wk-old euthymic C57BL/6 mice were Ab depleted as described above. Fourteen days after the first dose of Ab, mice received a heterotopic vascularized BALB/c cardiac allograft as previously described (20). Where indicated, mice received 5 × 106 irradiated donor-specific T-depleted spleen cells (DST) by i.v. injection on the day of transplantation as well a single dose of 200 μg of human CTLA4Ig (Bioexpress) on postoperative day 2. In some experiments, animals received graded doses of FACS-purified CD4+CD25+CD45RBlow T cells isolated from naive C57BL/6 mice. In others, mice were treated with the nondepleting anti-CD4 and anti-CD8 mAbs YTS177 and YTS105 (21) on days 0, 3, and 7 (1 and 0.2 mg/injection, respectively). Graft survival was monitored by palpation, with rejection defined as the absence of palpable contractions, and was verified by visual inspection of the graft after the animal was killed. For adoptive transfer experiments, recipients were thymectomized then rested for at least 1 wk after the procedure. The institutional animal use committee of University of Pennsylvania approved all procedures.
Foxp3 real-time PCR
Residual CD4 and CD8 cells surviving lymphocyte depletion demonstrate skewing toward memory phenotype
As noted above, our previous study found that the residual T cells in mice treated with suboptimal dosing of depleting Abs undergo homeostatic proliferation and are able to reject cardiac allografts despite treatment with costimulatory blockade. We hypothesized that this resistance was related to the wave of homeostatic proliferation occurring after subtotal lymphocyte depletion. Another explanation for tolerance resistance, however, was selective resistance of memory T cells (compared with naive and/or Tregs) to depletion (18) and/or preferential expansion by these cells during lymphopenia.
To address these questions, naive C57BL/6 mice were treated with two doses of depleting anti-CD4 and anti-CD8 mAbs on days 0, and 3. Consistent with our previous findings (10), this treatment resulted in an 80–90% reduction in total T cells isolated from LN and spleen when analyzed on day 10 (data not shown). Next, lymphocytes were isolated from mice at serial time points after mAb treatment and analyzed for surface phenotype. As shown in Fig. 1⇓, the population of residual lymphocytes present in the early aftermath of depletion was characterized by a significant decrease in CD62L expression as well as an increase in CD44 expression, consistent with the presence/emergence of cells with a memory phenotype as the dominant cell population after partial depletion. To exclude the effects of recent thymic emigrants, we repeated the above experiments in thymectomized mice and found similar results (data not shown), suggesting that for early time points, such as 7 and 10 days after the start of depletion, the effects of naive T cells being exported from the thymus contribute little to the overall T cell repertoire. Similar results were found in BALB/c mice, indicating that the differential depletion observed is not strain specific (data not shown).
Homeostatic expansion in the aftermath of lymphocyte depletion also contributes to skewing of the T cell repertoire
Although the appearance of more memory cells in the immediate aftermath of depletion supports the idea of differential depletion, it remained possible that the effects we observed were also due in part to homeostatic proliferation occurring very early after treatment with depleting Abs. To address this, we added BrdU to the drinking water of mice that had been previously thymectomized and treated with Ab depletion. A 6-day BrdU pulse was used, and mice were killed at various time points, ranging between 7 and 23 days after the beginning of depletion. We reasoned that if homeostatic proliferation were occurring contemporaneously with depletion, then we would rapidly observe increased BrdU uptake in the memory gate (CD44highCD62Llow) of depleted mice compared with the same gate in undepleted thymectomized mice. As shown in Fig. 2⇓, 7 days after the start of depletion, the profile of BrdU uptake was virtually identical among memory cells from depleted and undepleted mice despite a >80% reduction in total T cells after depletion. Similarly, there was virtually no BrdU incorporation at 7 days for naive cells (CD44lowCD62Lhigh) in either depleted or undepleted mice. These results were observed in both CD4 and CD8 populations. This indicates that 7 days represents a time point at which differential depletion has occurred, but homeostatic expansion has not yet become predominant.
As the time interval from depletion increases, BrdU uptake increases in all cell populations, but this is most prominent for memory cells. For CD4 cells, although there is up to a 10-fold increase (depending on time point) in BrdU incorporation in the naive cell population after depletion, the percentage of BrdU+ naive cells is always substantially less than the percentage of BrdU+ memory cells at the each time point. For CD8 cells, BrdU incorporation peaks at 10 days for both memory and naive cells, but falls rapidly for naive cells at later time points while remaining high in the memory cell fraction.
To explore further the relative efficiency of homeostatic proliferation among memory and naive CD4 T cells in the aftermath of depletion, we performed experiments in which CFSE-labeled purified memory or naive cells expressing the congenic markers Thy1.1 and CD45.1, respectively, were adoptively transferred into thymectomized mice that had received two doses of depleting Abs starting 10 days before the adoptive transfer. Mice were killed 5 days later, and LN and spleen cells were analyzed for CFSE dilution. As shown in Fig. 3⇓, recovered input CD4+Thy 1.1+ memory cells rapidly diluted their CFSE, with >90% of cells having divided more than seven times. In contrast, recovered input naive CD45.1+ demonstrated modest division, with roughly half the cells remaining undivided and the remainder dividing only once or twice. Only 3% of input naive cells showed complete dilution of their CFSE. Collectively, these data show that skewing of the T cell repertoire after depletion is initially due to relative sparing of memory cells, but is then amplified due to the enhanced postdepletion expansion of memory cells compared with naive cells.
Ab depletion does not markedly change the percentages of Tregs
We next asked what effect Ab depletion had on the pool of Tregs. Specifically we focused on the population of CD25+CD4+CD45RBlow cells, because these cells exist as a measurable minority population in naive mice and have been shown to promote long-term graft survival and tolerance in several transplantation models (22, 23, 24). Fig. 4⇓A shows that by cell surface staining criteria, among CD4 cells that survive subtotal lymphocyte depletion, the percentage that retain a CD25+CD45RBlow phenotype remains nearly equivalent to that in undepleted mice. We also assessed the effect of lymphocyte depletion on Foxp3 expression by both real-time semiquantitative PCR and direct intracellular staining. As shown is Fig. 4⇓B, residual CD4 cells surviving depletion expressed slightly more Foxp3 mRNA than undepleted CD4 cells. Importantly, the roughly 2-fold increase in Foxp3 expression was much less significant than the 27-fold difference between sorted CD4+25+ and CD4+25− cells. When depleted cells were subjected to Foxp3 intracellular staining, as shown in Fig. 4⇓C, roughly equal percentages, as a function of remaining CD4 cells, were seen compared with undepleted mice.
Homeostatic proliferation of CD25+CD4+CD45RBlow cells is slower than that of nonregulatory counterparts
We next asked whether CD25+CD4+CD45RBlow cells undergo homeostatic proliferation to a similar extent as other residual CD4 T cells. As in previous proliferation experiments, we used thymectomized mice to remove the confounding effects of thymic export on the proliferation of residual peripheral Tregs. As shown in Fig. 5⇓, regulatory cells (defined by cell surface staining) that survive depletion demonstrated a brief period of increased BrdU incorporation that peaked 6 days after depletion. Importantly, however, the degree of incorporation was markedly less than that of memory phenotype CD4+CD25− cells for each time point.
Exogenous Tregs can restore tolerance induction in Ab-depleted mice
Having shown that the lymphocyte pool is skewed in favor of memory cells in the aftermath of depletion and homeostatic proliferation without a comparable increase in CD4+CD25+ cells, we next asked whether providing exogenous additional regulatory T cells could restore the ability of costimulatory blockade to induce long-term graft survival. Naive C57BL/6 mice were partially lymphocyte depleted with anti-CD4 and anti-CD8 mAbs as described above. On day 14 after the start of depletion, the mice received a BALB/c allograft and were treated with DST and CTLA4Ig. As indicated, mice also received exogenous sorted CD4+CD25+CD45RBlow Tregs from naive C57BL/6 littermates. As shown in Fig. 6⇓ and consistent with our previous studies (10), treatment with DST and CTLA4Ig was not effective at inducing long-term graft survival in the context of partial lymphodepletion. However, the adoptive transfer of Tregs from naive mice significantly improved graft survival, with most mice exhibiting long-term graft function, although histologic assessment of function grafts on day 100 revealed significant parenchymal damage and scarring. Interestingly, very few Tregs were required; as few as 50,000 Tregs were as effective as 500,000, perhaps because of the known ability of Tregs to undergo homeostatic expansion or the possibility that they might become nonspecifically activated during that process.
Nondepleting mAbs restore tolerance in mice undergoing homeostatic proliferation
Waldmann et al. (25) have reported protocols in which the nondepleting CD4 and CD8 Abs, YTS177 and YTS105, can induce tolerance of cardiac allografts in the setting of a full MHC mismatch. More recently, the use of these Abs in combination with donor skin grafts has been shown to allow the induction of a population of Foxp3-expressing CD4+CD25+ T cells that infiltrate the allograft (26). Given these data and our finding that exogenous Tregs can prevent rejection in the Ab depletion model, we next asked whether the combination of YTS177 and YTS105 could prevent rejection in the context of homeostatic proliferation. C57BL/6 scid mice were reconstituted with 10 million unfractionated T cells from naive C57BL/6 mice. On the same day, these mice received a BALB/c cardiac allograft along with DST and CTLA4Ig or control human IgG. Mice then received YTS177 and YTS105 on days 0, 3, and 7. As shown in Fig. 7⇓, early treatment with this combination of nondepleting Abs along with DST and CTLA4Ig (but neither protocol alone) produced long-term graft survival. Histologic analysis of these grafts on day 150 revealed a variable picture, with half the grafts exhibiting minimal evidence of scarring and fibrosis, whereas others showed signs of damage (data not shown).
Collectively, these data indicate that a protocol of nondepleting mAbs can rescue the resistance to tolerance that is observed in the setting of homeostatic proliferation. Furthermore, the fact that indefinite graft survival was only seen after the addition of CTLA4Ig indicates that the combination of these nondepleting mAbs themselves is not tolerizing in the setting of homeostatic proliferation.
Nondepleting mAbs decrease the early wave of homeostatic proliferation
Recent work has shown that there are two distinct patterns of proliferation during lymphopenia: a slow pattern of T cell turnover that is dependent on IL-7 and characterized by one or two divisions in a 7- to 14-day period, and a faster wave of proliferation characterized by more than eight divisions in the same time frame (27). This faster wave of proliferation, also termed spontaneous proliferation, is characterized by much greater conversion of cells to the memory surface phenotype and an exaggerated increase in IFN-γ and IL-2 production. To test the effects of YTS177 and YTS105 on homeostatic proliferation, C57BL/6 scid mice were reconstituted with naive CFSE-labeled T cells from congenic Thy1.1+ C57BL/6 mice and then received two doses of YTS177, YTS105, or control Ab. As shown in Fig. 8⇓, both Abs were able to inhibit rapid lymphopenia-induced proliferation of their target cell populations. As anticipated, YTS177 had no effect on CD8 proliferation, and YTS105 had no effect on CD4 division (data not shown). Interestingly, we did not find significant differences in Foxp3 mRNA on CD4 LN and spleen cells recovered from animals treated with nondepleting Abs compared with animals given control IgG (data not shown).
Our previous work (10) had shown that homeostatic proliferation was a barrier to tolerance induction in both reconstituted scid mice and normal mice treated with T cell-depleting Abs. The later model is intriguing because of its similarities to induction therapy that is used to effect large-scale lymphodepletion in human solid organ transplant recipients. Although our initial report showed that treatment with depleting Abs induced a subsequent wave of homeostatic proliferation and the emergence of a repertoire skewed to memory cells, it did not define either the mechanisms underlying that event or ways to overcome it and induce tolerance.
The data in this study show that multiple mechanisms may be responsible for the emergence of memory T cells in the setting of lymphodepletion. First, we found that Ab depletion selectively targets naive cells and spares memory cells. This finding is consistent with a recent cohort study of human renal transplant recipients who received the potent lymphodepleting Ab CAMPATH 1-H (18). In the aftermath of depletion, the T cell repertoire was enriched toward CD4+ effector memory cells. Notably, without additional immunosuppression, all patients in this study developed acute rejection.
Our results do not indicate why selective depletion occurs. One possibility is that given their ability to migrate among lymphoid and nonlymphoid organs, memory cells experience lower concentrations of depleting Ab. Another largely untested possibility is that memory cells are intrinsically more resistant to depleting Abs, perhaps as a result of increased expression of decay-accelerating factor blocking complement-mediated lysis or of increased expression of antiapoptotic genes (18). A recent study of human lymphocytes treated with PHA showed that the small subset of cells surviving activation-induced cell death, presumably the cells that would become memory cells, also expressed increased levels of the antiapoptotic proteins c-FLIP and Bcl-xL (28). Whether such mechanisms exist for Ab depletion remains to be tested.
Second, although selective depletion plays a role in shaping the eventual T cell repertoire, we show in this study that memory cells undergo more rapid proliferation in the setting of lymphopenia. Hence, the lymphopenic aftermath of depletion offers memory cells a second selective advantage over regulatory and naive cells. The more rapid proliferation of memory cells is consistent with the results of a number of studies showing that they enter cell cycle more rapidly than naive cells after activation (29, 30). In addition, the known ability of memory cells to proliferate in response to low concentrations of Ag (31, 32, 33) and their relative lack of requirement for costimulatory signals (34) may also promote more rapid division in the lymphopenic environment. Finally, we and others (12, 13, 14, 15, 35) have previously found that naive cells are known to convert to memory-like cells after homeostatic expansion, and it is likely that this also contributes to repertoire skewing.
A number of investigators have shown that Tregs can prevent allograft rejection (23, 24, 36). Additional manipulations, such as use of the nondepleting mAb YTS177, can augment the expansion of a group of cells that are phenotypically similar to Tregs, express Foxp3, and prevent allograft rejection. Hence, it is possible that the resistance to tolerance induction seen after lymphodepletion might be related to a decrease in the relative percentage of T cells that have a regulatory phenotype. In contrast, because Tregs bear some similarities to memory T cells with regard to their surface phenotype and function (37, 38, 39), it was plausible that depletion might increase their numbers. We found that CD4+CD25+ Tregs, defined by surface phenotype, were not significantly increased as a percentage of remaining T cells after depletion and were only marginally increased when defined by Foxp3 staining. Although we show that Foxp3 expression does not significantly change after depletion, supporting the argument that lymphocyte depletion does not skew the regulatory compartment, our data have several caveats. First, we cannot say what effect depletion has on naturally occurring Tregs vs induced Tregs. Given that non-Tregs can convert in the periphery to Tregs, what we might be observing is the complex interplay of depletion of one Treg subset compensated by the outgrowth of another (40). Second, we cannot say what the long-term outcome of depletion will be on circulating regulatory populations. Hence, although we found a preserved percentage of regulatory cells 20 days after depletion, we also observed decreased BrdU uptake in the regulatory population compared with the memory-like population. It is possible that once homeostatic expansion has returned lymphocyte numbers to predepletion levels, which may take several months, the percentage of regulatory cells may be decreased compared with the nondepleted state.
If we view graft acceptance vs rejection as a temporal competition between effector cells and regulatory cells, then the ability of nondepleting Abs to allow for tolerance induction in the setting of lymphopenia may be due in part to their ability to blunt the rapid phase of homeostatic proliferation that is characterized by conversion of cells to the memory surface phenotype with accompanying induction of IFN-γ and IL-2 production (41). The fact that indefinite graft survival was only seen with the addition of CTLA4Ig indicates that the combination of these nondepleting mAbs themselves is not tolerizing in this setting and at these doses. Our finding that nondepleting Abs blunt proliferation does not necessarily run counter to the finding that these same Abs foster outgrowth of regulatory cells. If different populations of T cells compete for space and limiting growth factors, the suppression of one population could provide a selective advantage for another.
Our findings on regulatory cell proliferation are consistent with similar recent findings by investigators modeling homeostatic proliferation in RAG knockout mice (41). Although CD4+CD25+ cells were observed in the population of cells that had completely diluted CFSE, they were far fewer in number than memory phenotype cells. Given that depletion does not appear to favor regulatory cell outgrowth, the finding that exogenous Tregs can restore tolerance induction may be particularly important. Lakkis et al. (22) have shown that Ag-experienced, but not naive, CD4+CD25+ T cells can delay skin transplant rejection mediated by CD8 cells. In our model, naive Tregs were able to prevent rejection. This difference may simply reflect differences between the two models, the most notable being that in the heart transplant model, CD8 cells alone are insufficient to effect rejection. In contrast, it is tempting to speculate that lymphopenia itself and the resultant proliferation enhance the regulatory ability of CD4+CD25+ cells.
Peritransplant lymphodepletion is commonly used as inductive immunosuppression in clinical transplantation. However, based on our findings, such lymphodepletion may carry the risk of upsetting the natural balance of naive, memory, and regulatory T cells, and thus may prove a barrier to the induction of tolerance. We suggest this conclusion cautiously, because findings after a brief course of lymphodepletion in naive mice may not necessarily reflect long-term repopulation events after depletion in clinical transplantation. It should also be recognized that rare patients treated with depleting antilymphocyte sera have achieved tolerance. However, the degree of lymphodepletion achieved in those given antilymphocyte sera is not clear. Moreover, the finding of tolerance in a rare patient is still consistent with a concept of relative resistance to its induction.
Our data suggest that the adoptive transfer of Tregs, the use of strategies that spare regulatory cells or promote their outgrowth, and/or the addition of adjunctive agents (such as nondepleting T cell Abs) that blunt homeostatic proliferation in the residual nondepleted population may counterbalance the skewing of cells toward a memory phenotype and allow for long-term graft acceptance. Furthermore, because memory T cells have been implicated in tolerance resistance in a number of clinically relevant models, including sensitized hosts and heterologous immunity (16, 17, 42), as well as autoimmune disorders (43), these approaches may prove useful in transplantation and autoimmunity in a wide variety of clinical settings.
We thank the Flow Cytometry Core Facility at University of Pennsylvania for cell sorting. We thank Patrick Walsh and Somia Hickman Perdow for helpful discussion.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 Address correspondence and reprint requests to Dr. Laurence A. Turka, University of Pennsylvania, 700 Clinical Research Building, 415 Curie Boulevard, Philadelphia, PA 19104. E-mail address:
↵2 Abbreviations used in this paper: Treg, regulatory T cell; DST, donor-specific T-depleted spleen cell; Foxp3, Forkhead/winged helix transcription factor; LN, lymph node.
- Received December 6, 2005.
- Accepted February 2, 2006.
- Copyright © 2006 by The American Association of Immunologists