CD4+CD25+ Treg regulate the contribution of CD8+ T-cell subsets in repopulation of the lymphopenic environment


  • Afonso R. M. Almeida,

    Corresponding author
    1. Institute for Research in Biomedicine, Bellinzona, Switzerland
    2. Immunobiology Unit, Instituto de Medicina Molecular, Faculdade de Medicina de Lisboa, Lisboa, Portugal
    • Immunobiology Unit, Instituto de Medicina Molecular, Faculdade de Medicina de Lisboa, Av. Prof. Egas Moniz, Edifício Egas Moniz, 1649-028 Lisboa, Portugal Fax: +351-21-798-5142
    Search for more papers by this author
  • Ilja F. Ciernik,

    1. Radiation Oncology, Dessau Medical Center, Dessau, Germany
    2. Center for Clinical Research, Zurich University Hospital, Zurich, Switzerland
    Search for more papers by this author
  • Federica Sallusto,

    1. Institute for Research in Biomedicine, Bellinzona, Switzerland
    Search for more papers by this author
  • Antonio Lanzavecchia

    1. Institute for Research in Biomedicine, Bellinzona, Switzerland
    Search for more papers by this author


Peripheral T-cell expansion is of major relevance for immune function after lymphopenia. In order to promote regeneration, the process should result in a peripheral T-cell pool with a similar subpopulation structure as before lymphopenia. We investigated the repopulation of the CD8+ central-memory T cells (TCM) and effector-memory T cells (TEM) pools after adoptive transfer of sorted CD8+ T cells from naïve, TCM and TEM subsets into T-cell-deficient hosts. We show that the initial kinetics of expansion are distinct for each subset and that the contribution to the repopulation of the CD8+ T-cell pool by the progeny of each subset is not a mere function of its initial expansion. We demonstrate that CD4+CD25+ Treg play a major role in the repopulation of the CD8+ T-cell pool and that CD8+ T-cell subsets impact on each other. In the absence of CD4+CD25+ Treg, a small fraction of naïve CD8+ T cells strongly proliferates, correlating with further expansion and differentiation of co-expanding CD8+ T cells. CD4+CD25+ Treg suppress these responses and lead to controlled repopulation, contributing decisively to the maintenance of recovered TCM and TEM fractions, and leading to repopulation of each pool with progeny of its own kind.


Peripheral T-lymphocyte numbers are kept fairly constant throughout the life of an individual and from individual to individual while under homeostatic control 1. T cells can be divided in subpopulations that have specific functions in immune responses: naïve T cells provide the potential to respond to new antigens, whereas central-memory (TCM) and effector-memory (TEM) T cells can mount rapid and efficient secondary proliferative and effector responses to recall antigens 2–4. Thus, peripheral T-cell homeostasis includes a qualitative dimension, as maintenance of peripheral lymphocyte numbers should allow for maintenance of essential subpopulations 5.

One important feature of T-lymphocyte homeostasis is the ability to reconstitute peripheral T-cell numbers following severe depletion, such as those caused or induced by radiotherapy or chemotherapy 6–11, with or without contribution of newly produced cells from the thymus. However, although thymus-independent expansion of peripheral T cells leads to the recovery of T-cell numbers, the resulting cells are of ambiguous nature, as the process is accompanied by variable degrees of differentiation 9, 12. Moreover, the recovered T-cell pool has a restricted TCR repertoire, which is believed to be due to preferential proliferation of selected clones 5, 6, 9, 13.

Studies of lymphopenia-driven proliferation upon transfer into immunodeficient mice revealed that for both naïve CD4+ and CD8+ T cells, a fast-proliferating component, also defined as spontaneous, and a slow-proliferating component, also defined as homeostatic, exist, only with the latter being dependent on IL-7 14. Recent studies indicate that CD4+CD25+ Treg, essential in the control of lymphopenia-associated autoimmune diseases 15, 16, can inhibit the fast, spontaneous proliferation of naïve T cells without affecting the homeostatic component and prevent the development of CD4+ T-cell-mediated colitis, contributing to maintain TCR diversity 17, 18. However, the relative contribution of naïve, TCM and TEM cells to repopulation of the peripheral T-cell pool after lymphopenia and the role of CD4+CD25+ Treg in controlling lymphopenia-driven proliferation and differentiation of these populations have not been investigated.

In this study, we set out to determine the contribution of CD8+ T-lymphocyte subpopulations 4, 19 to the repopulation of lymphopenic CD3ε−/− mice (devoid of T cells but intact otherwise) 20. By transferring naïve, TCM and TEM cells carrying different congenic markers, we could determine the fate of progeny of each subset during repopulation when the three subsets were co-transferred into T-cell-deficient hosts. In addition, by varying the composition of the adoptively transferred cells, we revealed complex interactions that are involved in CD8+ T-cell repopulation, including a crucial role for CD4+CD25+ Treg.


Preferential expansion of equation image cells in lymphopenic hosts

To evaluate the contribution of CD8+ T-cell subsets in the repopulation of lymphopenic hosts, we set up an experimental system in which sorted naïve (CD44lowCD62Lhigh), TCM (CD44highCD62Lhigh) and TEM (CD44highCD62Llow) CD8+ T cells, that can be identified according to the expression of different congenic markers, are transferred, either alone or in combination, into CD3ε−/− mice. In this system, transferred cells undergo lymphopenia-driven expansion and progeny of naïve, TCM and TEM cells can be tracked at different time points by gating on the congenic markers (for experimental design, see Supporting Information Fig. 1A). In preliminary experiments, we evaluated the homing of naïve, TCM and TEM cells in different organs. After, 30 h, the recovery of cells was 10–15% of input and, although TEM cells were impaired in their ability to seed LN, the relative proportion of naïve, TCM and TEM cells in the spleen and in total secondary lymphoid organs was comparable to that measured before transfer (Supporting Information Fig. 1B).

We first transferred CD8+ T cells comprising equal proportions of naïve, TCM and TEM cells into CD3ε−/− mice, in order to directly compare the repopulation capacities of the three subsets while modeling the expansion in “older” individuals where subset distribution is skewed toward memory cells and evaluated the total number of recovered cells 8 wk after transfer. CD8+ T-cell recoveries showed a 38-fold overall expansion (Fig. 1A). Among the cells recovered, naïve progeny represented 27%, TCM progeny 58% and TEM 22%, corresponding to an increase of 27-, 64.8- and 22-folds, respectively. It should be noted that the fold expansion values are underestimated considering that the initial transfer efficiency is on the order of 10%. Thus, TCM showed the best repopulation capacity. We next transferred the CD8+ T cells in proportions biased toward naïve CD8+ T cells and with fewer TEM, thus modeling the situation in a young individual. In total, 8 wk later, CD8+ T-cell recoveries were in the order of 40×106, thus a 80-fold overall expansion (Fig. 1B), with naïve progeny having expanded 73-fold, TCM progeny 111-fold and TEM progeny 49-fold, thus all subsets expanded more in this situation. Among cells recovered, naïve progeny was still the majority, but only TCM progeny increased the fraction compared with transfer, suggesting that also in this situation TCM were best in repopulation, but that initial frequencies in the expanding population impact on the contribution from each subset.

Figure 1.

Contribution of CD8+ T-cell subsets in the repopulation of CD3ε−/− hosts. CD8+ T cells sorted according to cell-surface expression of CD44 and CD62L (Supporting Information Fig. 1) from donors differing in Ly5 congenic marker (Ly5.1, Ly5.2 and Ly5.1Ly5.2) were transferred into CD3ε−/− hosts. Relative proportions of each population transferred and recovered proportions of progeny from each population 8 wk after transfer of (A) 1.5×105 and (B) 5×105 total CD8+ T cells in secondary lymphoid organs (spleen+LN). Numbers in stacked bar graph are proportions in transferred mix or mean of respective progeny in four hosts. Numbers on top of bars are the absolute numbers transferred and recovered in the hosts. CD62L expression of progeny of each subset recovered (spleen+LN) 8 wk after transfer (bar graphs, mean±SEM) is also shown. (C) A mix of CD8+ T cells from the three subsets was CFSE stained and transferred into CD3ε−/− hosts (total CD8+ T cells transferred was 6×105). Hosts were sacrificed 3 or 7 days after transfer (two hosts/time point) and CFSE dilution profiles of progeny from each population recovered from the spleen are shown. Each experiment was repeated twice with similar results.

To ask whether the differential expansion of naïve, TCM and TEM cells correlated with differences in the kinetics of initial proliferation, we transferred a CFSE-stained CD8+ T-cell mix into CD3ε−/− hosts and analyzed the recovered cells on days 3 and 7 after transfer (Fig. 1C). TCM cells were the first to cycle, whereas naïve and TEM cells showed a slight delay, as shown by the CFSE profile on day 3. On day 7, almost all TEM cells were CFSE-negative, whereas TCM cells were homogenously CFSE-intermediate, and naïve T cells were composed of both CFSE-intermediate and CFSE-negative cells. In control experiments with CFSE-labeled T cells transferred into WT mice, naïve T cells did not divide, whereas very few TCM and TEM cells had undergone one cell division (Supporting Information Fig. 2A). This initial proliferation kinetics was reflected in the relative proportion of cells recovered on days 3 and 7 (Supporting Information Fig. 2B), but was not translated into contributions found at later time points (Fig. 1A and B). Taken together, these results indicate that TCM cells are the best contributors to long-term reconstitution and that factors other than initial homing and kinetics of proliferation contribute to the repopulation of lymphopenic hosts. These results also show that a small fraction of naïve T cells and the majority of TEM cells undergo rapid cell division in lymphopenic CD3ε−/− mice.

It is known that during an immune response or upon T-cell transfer, CD62L expression can be either lost or reacquired by memory CD8+ T cells 4. We therefore analyzed CD44 and CD62L expression on progeny from naïve, TCM and TEM cells recovered 8 wk after transfer into the CD3ε−/− mice (Fig. 1A and B). In all animals, total recovered cells uniformly expressed high levels of CD44, as expected (data not shown), whereas the expression of CD62L was heterogeneous but clearly biased toward CD62Llow when naïve CD8+ T cells represented a higher fraction of transferred T cells (Supporting Information Fig. 2C). When CD8+ T-cell subsets were transferred in equal proportions, 80% of naïve T-cell progeny lost CD62L expression, consistent with the acquisition of a TEM-like phenotype, whereas 73% of TCM progeny maintained high expression of CD62L, suggesting stability of the TCM phenotype. Finally, almost all TEM cell progeny retained a CD62Llow phenotype, indicating that in this condition there was no significant conversion from TEM to TCM-like cells. When the CD8+ T-cell subsets were transferred in fractions biased toward naïve T cells (Fig. 1B), however, progeny of all subsets was biased toward CD62low phenotype, thus changing initial frequencies impacted in the conversion rates of TCM to TEM phenotype.

Co-expanding equation image T cells impact on phenotype conversion

In order to investigate the intrinsic expansion capacity of the CD8+ T-cell subsets and the effects of co-expanding T cells in progeny of each subset, we transferred 5×104 CD8+ naïve, TCM and TEM cells alone or in combination into CD3ε−/− hosts and evaluated cell recovery and phenotype after 8 wk. The absolute numbers of cells recovered in the spleen and LN were comparable in mice that received either naïve or TCM cells, whereas the total number of cells recovered from mice that received only TEM precursors was significantly lower (Fig. 2A). The number of cells of naïve, TCM or TEM origin was comparable in mice in which these subsets were transferred alone or in combination, whereas the recovery of TCM progeny was slightly increased upon co-transfer with the other subsets. Thus, CD8+ T-cell recovery was higher when higher numbers were transferred, suggesting that CD8+ T-cell expansion failed to reach a homeostatic plateau in these conditions (Fig. 2A and B).

Figure 2.

Influence of other CD8+ T cells in the expansion and differentiation of progeny from each subset. (A) CD8+ T-cell recovery 8 wk after transfer of 5×104 precursors from each subset into CD3ε−/− hosts. A group of hosts received a mix of the same number of precursors from each subset (1.5×105 total CD8+ T cells transferred), differing in Ly5 congenic marker. Recovery in individual hosts and mean is shown. (B) Total CD8+ T-cell progeny from each subset 8 wk after expansion in CD3ε−/− hosts in the same experiment as in (A). Recovery in individual hosts and mean is shown. (C) Expression of CD127 and CD62L in progeny from each subpopulation recovered in the spleen of a representative host from each group. CD62L phenotype of recovered progeny from each subpopulation in the periphery (spleen+LN) of hosts transferred with each population alone or with a mix of the three subsets is summarized in the bar graphs (mean±SEM, n=9, 8, 7 and 8 for naïve only, TCM only, TEM only and mix, respectively). (D) Absolute numbers of CD44highCD62Llow progeny from each subset in hosts transferred with each subset (naïve (filled squares), TCM (open circles) and TEM (filled triangles)) alone or as a mix. Note increased recovery of TEM-phenotype progeny for both TCM and TEM. (E) Absolute numbers of CD44highCD62Lhigh progeny from each subset in hosts transferred with each subset (naïve (filled squares), TCM (open circles) and TEM (filled triangles)) alone or as a mix. A second independent experiment yielded equivalent results. *p<0.05, **p<0.01 and ***p<0.001, Student's t-test for unpaired samples.

Although the expansion capacity of naïve, TCM and TEM cells in lymphopenic hosts was little affected by transferring the cells alone or in combination, there was a change in the relative proportion of CD62Lhigh TCM-like and CD62Llow TEM-like cells recovered. In particular, the frequency (Fig. 2C) and/or absolute number (Fig. 2D) of CD62Llow cells were significantly higher in mice that had received the cell mix than in the hosts that had received each cell subset separately, whereas the fraction of TCM-phenotype TEM or naïve progeny was reduced. The absolute numbers of TCM-phenotype naïve or TEM progeny tended to be reduced upon co-transfer but were not statistically different in this experiment (Fig. 2E). Expression of CD44 and IL-2-IL-15Rβ did not change in the presence or absence of co-expanding populations (data not shown). These results suggest that the presence or absence of co-transferred CD8+ T cells impacts on the phenotype of progeny of the different CD8+ T-cell subsets.

equation image Treg impact mostly on the expansion and differentiation of naïve equation image T-cell precursors

In our first experiment (Fig. 1), we observed a major increase in CD8+ T-cell recoveries upon transfer of a higher fraction of naïve CD8+ T cells, but in order to vary the proportions we had transferred higher numbers of CD8+ T cells (5×105 compared with 1.5×105). We also observed that 8 wk after transfer, animals showed macroscopic signs of colitis (data not shown). Treg have been previously shown to regulate the expansion and differentiation of CD4+ T cells in lymphopenic hosts and to prevent the development of colitis 21–25. Therefore, we investigated the effects of cell numbers transferred and of CD4+CD25+ Treg, transferring into CD3ε−/− different total numbers of CD8+ T cells composed of naïve, TCM and TEM cells in a 1:1:0.5 ratio (Fig. 3A) in the absence or in presence of CD4+CD25+ Treg. Eight wk after transfer, we analyzed the recovered T-cell populations.

Figure 3.

CD4+CD25+ Treg play a selective suppressive role in the repopulation of the CD8+ T-cell pool. (A) Scheme of CD8+ T cells mixes transferred into the six groups of CD3ε−/− hosts. A mix of sorted CD8+ T cells from the three subsets (naïve, TCM and TEM) was prepared and then diluted 1/4 or 1/10 in order to transfer lower absolute numbers of cells (as indicated) at the same proportions in the three groups. CD4+ CD25+ Treg (105) were then added or not in the transfer. (B) Absolute numbers of CD8+ T cells are recovered in each of the groups (individual hosts and mean are shown). Note the reduced recovery in the higher transferred numbers group (mix) when in the presence of CD4+CD25+ Treg. Asterisks represent significance in the comparison of absolute numbers of hosts having the CD8+ T cells alone or with CD4+CD25+ Treg, with **p<0.01, Student's t-test for unpaired samples. (C) Contribution to the total recovered CD8+CD3+ pools in hosts transferred with the CD8+ T-cell mix alone (top) or co-transferred with CD4+CD25+ Treg (down). Numbers in stacked bar graphs are proportions in the transferred mix or mean contribution of respective progeny to the total recovered CD8+CD3+ cells. (D) CD62L expression in recovered CD8+CD3+ T cells from the spleen of a representative host from each group. (E) Contributions of each population to the recovered CD44highCD62Lhigh and CD44highCD62Llow CD8+CD3+ T cells in each group of hosts. Numbers in stacked bar graphs are mean contribution of progeny from each subpopulation (error bars are SEM, n=4). A second independent experiment with four animals/group yielded equivalent results.

In the absence of Treg, the total number of cells recovered was roughly proportional to the input cell number (25×106, 7×106 and 4×106 in mice receiving 5×105, 1.25×105 and 0.5×105 cells, respectively). On the contrary, in mice co-transferred with Treg, the recovery of total CD8+ T-cell numbers was lower compared with the absence of Treg and comparable at all T-cell inputs (3×106, 5×106 and 2.5×106). Interestingly, the contribution of naïve T-cell progeny to the recovered population, 36, 18 and 12% in the absence of Treg, was decreased to 8, 8 and 9% in the presence of Treg for cell inputs of 1 (mix), 1/4 and 1/10, respectively (Fig. 3C). Treg also dramatically decreased the percentage of CD62Llow cells recovered (Fig. 3D). Thus, in the absence of Treg, the frequency of CD62low cells was dependent on the input (75, 45 and 11%, for cell inputs of 1 (mix), 1/4 and 1/10, respectively), whereas in the presence of Treg it was virtually constant (10, 13 and 14%).

We next determined the contributions of each subset to the recovered CD44highCD62Lhigh TCM-phenotype and CD44highCD62Llow TEM-phenotype CD8+ T-cell pools (Fig. 3E). Although TCM cells were the main contributors to the recovered CD62Lhigh T cells in all groups, this was particularly so in hosts receiving lower inputs of cell numbers or in groups where CD4+CD25+ Treg were included, whereby more than 90% of recovered CD62Lhigh T cells were of TCM origin. CD62Llow T cells were derived from all three subsets, but the decrease in their recovery observed upon transfer of higher cell numbers in the presence of CD4+CD25+ Treg was mostly the result of the decreased contribution of naïve progeny and to a lesser extent of TCM progeny (Fig. 3E and Supporting Information Fig. 3), resulting in a higher contribution from cells of TEM origin (Fig. 3E). Indeed, upon transfer of CD8+ T cells at even proportions of the different subsets in the presence of CD4+CD25+ Treg, the vast majority of recovered CD62Lhigh and CD62Llow CD8+ T cells were of TCM and TEM origin, respectively (Supporting Information Fig. 3D).

Taken together, these results suggest an important role for CD4+CD25+ Treg in maintaining the TCM and TEM phenotype in CD8+ memory T cells undergoing lymphopenia-induced expansion. Thus, in the absence of Treg, the total number recovered, the contribution of each population and the phenotypic conversion of expanded CD8+ T cells are dependent on total cell input. On the contrary, in the presence of Treg, the process of reconstitution reaches the same endpoints both in terms of cell number and in terms of phenotype, irrespective of the initial T-cell input. In particular, Treg limit both the expansion and the differentiation of naïve T cells and, to a lower extent, the expansion of TEM cells. Examination of absolute cell numbers confirmed that Treg suppress naïve and TEM cell expansion when these populations were transferred at high numbers, whereas there was only a minimal effect on the expansion of TCM cells (Supporting Information Fig. 4).

equation image Treg suppress the fast-proliferating cohort of naïve equation image T cells

Considering the early expansion kinetics of the naïve T-cell pool (Fig. 1) and the effect of CD4+CD25+ Treg in regulating the expansion of naïve CD8+ T cells in lymphopenic hosts, we hypothesized a selective role of CD4+CD25+ Treg in the suppression of the fast-proliferating cohort of naïve CD8+ T cells. To test this assumption, we transferred each CFSE-labeled population alone or together into CD3ε−/− hosts, in the presence or absence of CD4+CD25+ Treg (Fig. 4). We evaluated the expansion kinetics at early time points (days 7 and 14 after transfer) and the repopulation efficiency at a later time point (8 wk).

Figure 4.

CD4+CD25+ Treg suppress the fast proliferating cohort of naïve progeny. CD8+ T cells from different subpopulations were sorted and stained with CFSE prior to transfer (2×105 for naïve and TCM, 105 for TEM) into CD3ε−/− hosts either alone or together as a CD8+ T-cell population mix with or without CD4+CD25+ Treg. Two hosts from each group were sacrificed 7 or 14 days after transfer and CFSE dilution patterns, absolute numbers and contributions from each population evaluated. (A) CFSE profiles in CD8+CD3+ T cells from each original cell population recovered at the indicated time points in the spleen. Numbers shown in the naïve origin histograms correspond to naïve progeny and are the mean values for the fast and slow proliferating cohorts from the two animals/group. (B) Contributions from each subpopulation were determined at days 7 and 14 or 8 wk (n=3) after transfer and are presented in the stacked bar graphs. Numbers in graphs are mean contribution of the progeny from each subpopulation to the total recovered donor-derived CD8+CD3+ T-cell pool (error bars are SEM). Results shown are representative from two independent experiments.

At 7 or 14 days after transfer (Fig. 4A), no change in the CFSE dilution profiles of each subpopulation was observed when co-transferred with the other CD8+ T-cell subpopulations. However, when CD4+CD25+ Treg were also transferred, a selective effect in the fast-proliferating cohort of naïve progeny was apparent by day 7 after transfer and this was more pronounced at day 14 (Fig. 4A), whereas no effect of the CD4+CD25+ Treg co-transfer was observed at these time points in the CFSE dilution profiles of both TCM and TEM progeny (Fig. 4A). As a consequence, the contribution of naïve progeny was clearly lower at day 14 in hosts co-transferred with CD4+CD25+ Treg co-transfer (10% compared with 32% in the presence or absence of CD4+CD25+ Treg, respectively) and the proportions were already similar to the ones found at 8 wk (Fig. 4B). The data imply that CD4+CD25+ Treg play a crucial role at early time points and that the remaining CD8+ T cells from each subset accumulate at similar rates. In accordance, recovery of CD8+ T cells in the “slow” fraction of naïve progeny was not affected (Fig. 4A, see numbers in histograms), implying that early CD4+CD25+ Treg effects are specifically targeting the fast-proliferating cohort of naïve progeny.

equation image Treg, competition and bystander help play different roles in equation image T-cell repopulation

Our results suggest that the increase in the proportion of TEM-phenotype progeny from TCM and TEM CD8+ T cells was secondary to the expansion of the fast-proliferating cohort of naïve CD8+ T cells. This could be an explanation for the observation that CD4+CD25+ Treg had no direct impact in TCM and TEM progeny early on (Fig. 4), but both TCM and TEM progeny were affected at late time points. We analyzed the phenotype of progeny from each subset recovered at 8 wk after transfer to evaluate the impact at later time points of co-expanding CD8+ T cells and CD4+CD25+ Treg on absolute numbers of TEM-phenotype and TCM-phenotype progeny from each origin (Fig. 5).

Figure 5.

Influence of other CD8+ T-cell subpopulations and CD4+CD25+ Treg on the expansion and repopulation of CD8+ T cells. (A) CD8+ T cells from different subpopulations were sorted and transferred (2×105 for naïve and TCM, 105 for TEM) into CD3ε−/− either alone or together as a CD8+ T-cell population mix with or without CD4+CD25+ Treg (same experiment as in Fig. 4). Graphs show total CD8+ T-cell recovery (top), CD62Lhigh progeny (middle) and CD62Llow progeny (bottom) (individual hosts and mean are shown) from naïve (filled squares), TCM (open circles) and TEM (filled triangles) origin in each of the groups 8 wk after transfer. (B) Role of CD4+CD25+ Treg in the recovery of CD62Llow cells from TCM or TEM progeny. Sorted TCM CD8+ T cells (3×105) (left) or TEM CD8+ T cells (1.5×105) (right) were transferred alone or co-transferred with 105 CD4+CD25+ Treg. The graphs show the number of CD44hghCD62Llow progeny from TCM (left) or TEM (right) recovered (individual hosts and mean are shown) in the secondary lymphoid organs 8 wk after transfer. Results shown are from a representative experiment out of the two (A and B, right) or three (B, left) independent experiments yielding similar results. *p<0.05, **p<0.01 and ***p<0.001, Student's t-test for unpaired samples.

Evaluation of total CD8+ T-cell recovery yielded no differences in the total numbers of progeny from each subset when transferred alone or in combination with the other CD8+ T-cell subsets, whereas only naïve and TEM progeny were reduced when CD4+CD25+ Treg were co-transferred (Fig. 5A, top). CD44highCD62Lhigh TCM-phenotype progeny from all subsets were reduced in hosts transferred with the CD8+ T-cell mix compared with hosts having received each subset alone (Fig. 5A, middle), suggesting competition for the TCM niche. When CD4+CD25+ Treg were included, no further reduction in the recovery of TCM-phenotype progeny from TCM or TEM origin was observed and only naïve TCM-phenotype progeny was further reduced. On the contrary, we could observe an increase in CD44highCD62Llow TEM-phenotype TCM progeny when CD8+ T-cell subsets were co-transferred as shown in Fig. 2, although variation prevented the differences to be statistically significant in this experiment. When CD4+CD25+ Treg were included, a strong reduction in the recovery of CD62Llow TEM-phenotype progeny from all subsets was observed, to numbers below those measured when the populations were transferred alone, further suggesting a distinct direct effect of CD4+CD25+ Treg on all subsets, in addition to the abrogation of bystander effects (Fig. 5A, bottom panel). In order to confirm this, we transferred either TCM or TEM CD8+ T cells alone or with CD4+CD25+ Treg and evaluated CD44highCD62Llow TEM-phenotype recovery 8 wk after transfer (Fig. 5). Indeed, CD4+CD25+ Treg co-transfer had also a direct effect in the recovery of CD62Llow progeny from TCM or TEM CD8+ T cells (Fig. 5B).


The regeneration of the peripheral T-cell pool after lymphopenia should give rise to a peripheral pool with equal quantitative and qualitative characteristics as before the onset of lymphopenia, i.e. re-establish the steady state. Although recovery of the peripheral naïve T-cell pool is dependent on thymic output 12, 26, the absolute numbers of memory phenotype T cells can recover to a large extent through peripheral expansion 6, 27. In this study, we show that progeny from all major CD8+ T-cell subsets (naïve, TCM and TEM cells) is found in the repopulated CD8+ T-cell pool after adoptive transfer into T-cell-deficient hosts (Fig. 1). However, the contribution from each subset does not reflect the initial input, as cells of TCM origin are over-represented in the recovered population of secondary lymphoid organs. Although TCM and TEM cells migrated less efficiently to LN compared with naïve T cells, all three subsets could be found in the spleen at early time points after transfer in similar proportions as the input cells. As a large majority of recovered CD8+ T cells were found in the spleen, we conclude that the overall seeding capacity of the three subsets is similar. Thus, the different repopulation capacities of naïve, TCM and TEM cells likely depend on differences in proliferation and/or survival. It will be interesting to verify whether the repopulation of CD8+ T-cell pools in nonlymphoid organs follows the same trend or is in contrast biased to TEM progeny, and thus implying that part of the poor repopulation capacity of TEM CD8+ T cells is related to migration into tissues.

CFSE experiments (Fig. 1) revealed that each CD8+ T-cell subset shows characteristic kinetics in their initial proliferation, which may reflect not only differences in sensitivities to activation and proliferation signals 9, 28, but also differences in TCR usage and repertoire diversity 13. It is conceivable that the fast-proliferating population present in the naïve CD8+ T-cell pool contains T-cell clones responding to antigenic stimulation and undergoing full-blown responses (see below). The fact that very few TCM CD8+ T cells proliferate with similar fast kinetics suggests that cognate antigens are not present in the lymphopenic hosts and that the transferred cells have a restricted repertoire of expanded clones. However, TCM cells appear to have a higher capacity to respond to homeostatic signals than naïve T cells, which is consistent with the finding that these cells have a higher expression of CD122 (IL-15Rβ) 29. The very fast proliferation kinetics of TEM cells is puzzling. Although we cannot exclude that the proliferation of these cells is antigen dependent, we favor the possibility that homeostatic signals act strongly on TEM cells, as transferred TEM cells should also have a restricted repertoire of expanded clones and have been characterized precisely as performing poor proliferative responses to antigen 4. TEM cells have been found to rapidly respond and differentiate upon antigenic stimulation 4, and this characteristic can thus be maintained in response to lymphopenia. In spite of the high proliferation, the contribution of TEM cells to T-cell reconstitution in lymphopenic hosts is low, suggesting that proliferation leads to terminal differentiation and cell death. These results highlight the differences between the early kinetics in response to lymphopenia and a productive contribution to repopulation, and suggest that a note of caution should be added when interpreting data from CFSE dilution experiments. Indeed, more than the capacity to proliferate, repopulation will be determined by the maintenance of progeny from a given subset. Thus, later events occurring after the initial period of proliferation may be of greater relevance to the regeneration of the T-cell pool after lymphopenia.

Previous studies have shown that lymphopenia-driven proliferation leads to changes in phenotypic markers and, in the case of naïve T cells, to the differentiation into memory-like cells 9, 12. Our data provide interesting information on how this phenotypic conversion is regulated. We show that the proportion of CD62Llow TEM-like cells in the recovered population was much higher in mice co-transferred with the three CD8+ T-cell subsets at higher numbers (Fig. 3) and when naïve CD8+ T cells were transferred at higher proportion (Fig. 1B), leading to the increase in the conversion rate of TCM cells toward CD62Llow or in the degree of expansion of CD62Llow progeny of TCM. Thus, co-transferring the CD8+ T-cell subsets correlates to higher TEM recoveries. On the contrary, TCM progeny of naïve and TEM tends to be lower upon co-transfer with the other subsets, suggesting competition with TCM progeny. Although our results (Fig. 2 and 5) are in apparent contradiction, the lower absolute numbers transferred in the experiment shown in Fig. 2 compared with the experiment in Fig. 5 would be expected to result in less apparent competition. This points to different mechanisms in the homeostatic control of TEM and TCM pools: the differentiation of naïve cells and bystander co-expanding TCM progeny to TEM-like cells may be favored by the release of pro-inflammatory cytokines by the fraction of naïve CD8+ T cells that rapidly proliferate and differentiate into CD127low effector cells 30, whereas competition for a limited TCM niche (possibly defined by IL-7 and IL-15 levels) may lead to limited naïve and TEM conversion to TCM.

This idea is further supported by the experiments including co-transferred CD4+CD25+ Treg. At high-input cell numbers, Treg suppressed the fast-proliferating cohort of naïve T cells while having only a marginal effect on the proliferation of TCM and TEM cells and the rate of conversion to TEM-like cells was much lower than in the absence of Treg (Fig. 3–5). These results extend CD4+CD25+ Treg action in the control of CD8+ T-cell-mediated autoimmunity and are in accordance with data published by others while this article was in preparation 17, 18, 25, 31. Indeed, in the absence of CD4+CD25+ Treg, recipients of higher numbers of CD8+ T cells develop macroscopic symptoms of colitis (data not shown). Thus, although CD4+CD25+ Treg are able to prevent the expansion of “auto-reactive” CD8+ T-cell clones from the initial expansion phase, in their absence, these cells will expand and provide bystander help to other CD8+ T cells leading to further expansion and/or differentiation of TEM-phenotype CD8+ T cells. At later time points, this would result in increased progeny from all subpopulations in the host as we observed when higher numbers and a higher fraction of naïve CD8+ T cells were transferred (Fig. 1B) and tend to mask the origin of the pathogenic response, which can be clearly observed at early time points. The fast-proliferating naïve clones are present at low frequencies (1 in 5–10 000, considering 10–15% homing efficiency) and are thus absent in recipients transferred with lower absolute numbers of CD8+ T cells where no CD4+CD25+ Treg effect was measured (Fig. 3). This may explain the reason why we had previously failed to see an effect of CD4+CD25+ Treg co-transfer on CD8+ T-cell expansion 24. Indeed, in recipients of low absolute numbers of the CD8+ T-cell mix or in any of the recipients co-transferred with CD4+CD25+ Treg/CD8+ T cells, no macroscopic symptoms of colitis were observed.

The goal of this study was to determine the relative contributions of precursor subsets to the repopulated memory phenotype CD8+ T-cell pools. Although it was clear from our initial experiments that each population can give rise to progeny of “converted” phenotype, our results show that in order to predict the endpoint of CD8+ T-cell expansion in lymphopenic hosts three key aspects must be taken into consideration: first, the initial subset composition of the expanding CD8+ T cells, implying that age groups should be considered when predicting the result of LDP; second, the absolute number of CD8+ T cells expanding (or the degree of lymphopenia), implying that for different lymphopenic states distinct mechanisms will prevail; third, the presence or absence of CD4+CD25+ Treg which we show in this study that can regulate the process of repopulation. Thus, in the absence of CD4+CD25+ Treg, and if sufficient numbers are available, bystander helper effects would result in additional CD44highCD62Llow CD8+ T cells from all origins, masking a lineage stability trend. Indeed, in the presence of CD4+CD25+ Treg, the majority of recovered TCM and TEM cells are progeny of their initial precursor, i.e. TCM and TEM precursors, respectively. Our results strongly suggest that in more physiological situations (thus in the presence of CD4+CD25+ Treg) memory CD8+ T-cell pools are thus regenerated mostly through the direct progeny of each subset rather than through conversion of one subpopulation to another. In accordance, in parallel studies where the contribution of CD8+ T-cell subset to reconstitution issues was investigated in irradiated hosts where radio-resistant CD4+CD25+ Treg persist (Almeida et al., unpublished data) we found that almost all donor-derived CD8+ T cells were TCM progeny, and almost all of the recovered CD8+ T cells display a TCM phenotype.

Interestingly, we found that a greater fraction of TEM repopulation is derived through TCM progeny rather than the repopulation of TCM resulting from TEM expansion (Fig. 5), suggesting that differentiation is more likely to occur in one but not the other direction. Indeed, the capacity of either naïve or TEM progeny to regenerate the TCM pool was poor and additionally impaired in co-expansion through competition with TCM progeny. It remains to be established whether in the presence of other CD4+ T-cell subsets competition with CD4+ T cells promotes changes in CD8+ T-cell repopulation, however, CD4+CD25+ Treg seem to promote the same regulation in the CD4+ T-cell pool, as they regulate CD4+CD25-CD45RBhigh naïve T-cell expansion but seem to impact little on the expansion of CD25CD45RBlow memory CD4+ T cells 24, 32. Further studies will address these aspects and the molecular mechanisms involved, in order to modulate repopulation after lymphopenic states.

In conclusion, in this study we show that the capacity to attain homeostatic equilibrium, the contribution to a given subpopulation after expansion and the direction of differentiation of a CD8+ T-cell subset will be determined by the “social” context, i.e. the presence of other co-expanding lymphocytes, in particular CD4+CD25+ Treg. Considering that the closest to regeneration would be the reconstitution of the peripheral CD8+ T-cell pools with the same specificities in the same pools, the nearest possible would be to regenerate each with progeny from its own remaining cells. We found that in the presence of CD4+CD25+ Treg, this is the general tendency. We identified CD4+CD25+ Treg-mediated suppression as a crucial component in CD8+ T-cell repopulation, whereas our results strongly suggest that bystander expansion and competition are important mechanisms acting in the peripheral expansion of CD8+ T cells following lymphopenia in health and disease. This knowledge improves our understanding of T-cell regeneration, providing help in predicting its success or failure. More directly, it provides a basis for the design of adoptive transfer immunotherapy strategies, which should take into account the interactions between the progeny of different precursor cells and also with host lymphocytes.

Materials and methods


C57Bl/6.Ly5.2 animals were obtained from Harlan (Milano, Italy) and congenic BA (Ly5.2Thy1.1), Hz (Ly5.1-Thy1.1), C57Bl/6Ly5.1, heterozygous F1 (BA×HZ), expressing Ly5.1-Ly5.2 and B6.CD3ε−/−20 were bred in our animal facilities. Animals were sex and age matched. Mice were treated in accordance with the guidelines of the Swiss Federal Veterinary Office and experiments were approved by the Dipartamento della Sanità e Socialità. Canton of Ticino, Switzerland.

Antibodies and flow cytometry

The following monoclonal antibodies were used: anti-CD3ε (145-2C11), and anti-CD122 (TM-β1) from BD Biosciences (San Diego, CA, USA), anti-CD4 (L3T4/RM4-5), anti-CD25 (7D4), anti-CD45RB (16A), anti-CD44 (IM781), anti-CD62L (MEL14), anti-Ly5.1 (A20), anti-Ly5.2 (104), anti-CD8a (53-6.7) and anti-CD127 (A7R34) from eBioscience (San Diego, CA, USA). Cell-surface four-color staining was preformed with the appropriate combinations of FITC, PE, PerCP, PE-Cy7, APC, APC-Cy7, APC-Alexa750, Biotin and Pacific Blue-conjugated antibodies. Streptavidin conjugated to APC-Cy7 (eBioscience) or PerCP (BD Biosciences) was used in the secondary labeling of biotin-conjugated antibodies. Dead cells were excluded during analysis according to their light-scattering characteristics. All acquisitions were performed with a FACSCanto or a FACSCanto II (Becton Dickinson, San Jose, CA, USA) interfaced to Facs Diva software. Analysis was performed with Flowjo software (TreeStar). CFSE staining was performed as described previously 33. Briefly, the sorted CD8+ T cells were incubated 8 min at 37°C with CFSE (5 μM).

Cell sorting and cell transfers

Spleen cells from Ly5.2, Ly5.1 and Ly5.1-Ly5.2 donor mice were enriched for CD8+ T cells by positive selection with anti-FITC magnetic beads (Myltenyi Biotec, Germany) after staining cells with anti-CD8α-FITC. Cells were then labeled with the appropriate combinations of anti-CD44, anti-CD62L and anti-CD4 antibodies and sorted on a FACSAria (BD Biosciences) as CD4CD8+CD44CD62Lhigh (naïve), CD4CD8+CD44highCD62Lhigh (TCM) and CD4CD8+CD44highCD62Llow (TEM), as shown in Supporting Information Fig. 1. To isolate CD4+CD25+ Treg, LN cells were first labeled with anti-CD4-FITC antibodies and the cell suspension was enriched for CD4+ T cells by magnetic sorting using anti-FITC magnetic beads (Myltenyi Biotec). The CD4+-enriched suspension was then labeled with anti-CD25-PE (Southern-Biotechnologies, clone 7D4) The purity of the sorted CD45RBhighCD25CD4+ and CD45RBlowCD25+CD4+ populations varied from 96 to 99.9%.

Intact nonirradiated B6 or B6.CD3ε−/− hosts were injected i.v. with purified CD8+ and CD4+ T-cell populations alone or mixed as described in the text and figure legends. In each co-transfer experiment, the proportions of the transferred populations were verified by FACS analysis prior to transfer. Host mice were sacrificed at different time intervals after cell transfer, as indicated. Spleen, inguinal and mesenteric LN cell suspensions were prepared and the number and phenotype of the cells from each donor population evaluated using antibodies to Ly5.1 and Ly5.2. The total peripheral T cells shown in the results represent the number of cells recovered in the host's spleen added to twice the number of cells recovered from the host's inguinal and mesenteric LN.

Statistical analysis

Means were compared using unpaired Students' t-test. In the case that variances of two samples were considerably different, the data were log transformed and compared using unpaired t-test or Welsh's correction applied to the original data if variances were still different after transformation. Data were considered significantly different at p<0.05. *p<0.05, **p<0.01 and ***p<0.001.


The authors thank David Jarrossay for cell sorting; Enrica Mira Catò and Luana Perlini for technical assistance; António Freitas for discussions and reading the manuscript; Dior Kingston for help preparing the manuscript and Andrea Reboldi, Giorgio Napolitani, Dirk Baumjohann and Henrique Veiga-Fernandes for discussions. This work was in part supported by the Swiss National Science Foundation (Grants no. 31-126027 to A. L. and 31-116440 to F. S.) and the Cancer League of Zurich, Switzerland to I. F. C. The Institute for Research in Biomedicine is supported by the Helmut Horten Foundation.

Conflict of interest: The authors declare no financial or commercial conflict of interest.