Report

Regulatory T Cells Increase the Avidity of Primary CD8+ T Cell Responses and Promote Memory

See allHide authors and affiliations

Science  26 Oct 2012:
Vol. 338, Issue 6106, pp. 532-536
DOI: 10.1126/science.1227049

Abstract

Although regulatory T cells (Tregs) are known to suppress self-reactive autoimmune responses, their role during T cell responses to nonself antigens is not well understood. We show that Tregs play a critical role during the priming of immune responses in mice. Treg depletion induced the activation and expansion of a population of low-avidity CD8+ T cells because of overproduction of CCL-3/4/5 chemokines, which stabilized the interactions between antigen-presenting dendritic cells and low-avidity T cells. In the absence of Tregs, the avidity of the primary immune response was impaired, which resulted in reduced memory to Listeria monocytogenes. These results suggest that Tregs are important regulators of the homeostasis of CD8+ T cell priming and play a critical role in the induction of high-avidity primary responses and effective memory.

In the absence of Foxp3+ regulatory T cells (Tregs), multiple organ autoimmune pathologies arise, which lead to death in both humans and mice (1). Tregs suppress autoreactive T cells through multiple effector mechanisms, acting both during priming in lymph nodes and during the effector phases of immune and inflammatory responses (15). It is intriguing that, in healthy individuals, Treg-mediated suppression does not compromise T cell responses to infectious, nonself antigens. Previous reports propose that Tregs not only inhibit immune responses to nonself antigens but also may contribute to clearance of viral or parasitic infections (6, 7). How Tregs contribute to the priming of T lymphocytes to nonself antigens remains unclear.

To investigate the role of Tregs in CD8+ T cell responses to nonself antigens, we used mice expressing the human diphtheria toxin receptor under the control of the Foxp3 promoter (DEREG mice). After two diphtheria toxin (DT) injections (8), we confirmed Treg depletion and the absence of any detectable polyclonal T cell activation or changes in dendritic cell (DC) number and activation (9).

We first analyzed the response to the H-Y male-specific histocompatibility antigen Uty in female mice. We immunized Treg-depleted or control female mice with male splenocytes and measured the response against a peptide derived from the Uty antigen by using Db-Uty multimers. Immunization of control females induced a uniform population of CD8+ T cells strongly labeled by the multimer (Fig. 1A). Depletion of Tregs during priming (subgroup G1, received DT on days –1, 0, 5, and 6) (fig. S1A) resulted in the expansion in all mice analyzed (n = 20) of a distinct population of CD8+ T cells that bound small amounts of multimers, accompanied by an increase in the total numbers of multimer-positive cells (Fig. 1, A to C). The expression of CD8 was not modified (fig. S1B). In some Treg-depleted mice, the high multimer–binding CD8+ T cell population disappeared, most likely because of competition or differences in the T cell receptor (TCR) repertoires. The intensity of multimer binding reflects the affinity of the TCR for major histocompatibility complex (MHC)–peptide complexes (1015). The ratio between the mean fluorescent intensity (MFI) for multimer to the MFI for TCR expression (relative affinity) was decreased in the absence of Tregs, as compared with control littermates (Fig. 1D and fig. S1C). A better estimate of the avidity of the polyclonal response can be obtained using a multimer dilution assay (12, 15, 16), where the avidity of the response is estimated using the half maximum response concentration (EC50). The Db-Uty multimer dilution assay showed a clear decrease in the avidity of the Uty-specific CD8+ cells upon depletion of Tregs (Fig. 1, E and F, and fig. S1D).

Fig. 1

Treg depletion impairs the affinity of H-Y–specific CD8+ T cell responses during naive T cell priming. Flow cytometric analysis of Uty-specific CD8+ T cells in the spleen of female littermates and DEREG mice immunized with 5 × 106 male splenocytes intraperitoneally (i.p.) that received DT injections on days –1, 0, 5, and 6. Primary CD8+ T cell responses were measured by Db-Uty–multimer staining 12 days after immunization. (A) Representative CD8+-gated plots are shown. (B) Frequency of Db-Uty multimer–positive among CD8+ T cells in the spleen. L: littermate; D: DEREG; PBS, phosphate-buffered saline. (C) The numbers of high-avidity and low-avidity antigen-specific cells were determined by Db-Uty–multimer staining. (D) The relative affinity (multimer/TCR MFI ratio) of Db-Uty–specific CD8+ T cells. (E) Direct ex vivo multimer-dilution assay. The percentage of Db-Uty multimer–positive cells are normalized to the number of Db-Uty multimer–positive cells at the highest multimer concentration; individual mice (black line) and the mean value for each group (red line) are represented. (F) For each mouse, the data obtained in (E) were analyzed to fit to sigmoid dose-response curves and the EC50 value was calculated. Results from at least three independent experiments are shown. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.

To investigate whether the impact of Tregs on CD8+ T cell avidity operates during the early phases of T cell activation (i.e., priming) or later during T cell expansion, we delayed the depletion of Tregs (G2: DT on days 3, 4, 9, 10; G3: DT on days 5, 6, and 11) (fig. S1A). Delayed Treg depletion still caused an increase in the percentage of Db-Uty+ T cells, indicating that, in all cases, Tregs limit T cell expansion (fig. S2A). Decreased T cell avidity, by contrast, was only observed when the depletion of Tregs was performed during priming (fig. S2, B to D). We conclude that the presence of Tregs during T cell priming to a nonself antigen increases the affinity of the CD8+ T cell response, most likely by inhibiting the priming of T cells bearing low-avidity antigen-specific TCRs.

To generalize this observation to other antigens, we next analyzed a CD8+ T cell response to the OVA peptide 257–264 (SIINFEKL, N4) after immunization with peptide-loaded lipopolysaccharide (LPS)–treated DCs and obtained similar results (fig. S3, A to D). In spite of the increase in total Kb-N4 multimer–positive cells after priming in Treg-depleted mice (fig. S3B), the number or relative affinity of antigen-specific CD8+ memory T cells after challenge (which is increased compared with the avidity of the primary cells) was not enhanced in DEREG mice deprived of Tregs during priming (fig. S3, B and D). Indeed, the ratio between the mean number of memory and primary multimer–positive cells was decreased upon Treg depletion during priming (fig. S3C), which indicated that the numerous low-avidity OVA-specific T cells that proliferate in the absence of Tregs do not give rise to long-lived memory cells.

Because Treg depletion can perturb the environment, which could indirectly affect T cell priming, we decided to test if injection of isolated antigen-specific Tregs would increase the avidity of a polyclonal T cell response. We isolated Dby-specific, Foxp3, enhanced green fluorescent protein–negative (EGFP) T helper cells (TH) or Foxp3+ EGFP+ Tregs from Foxp3-EGFP CD4+ TCR-transgenic Marilyn mice. We injected these cells into wild-type hosts previously immunized with (N4 + Dby)-loaded DCs. The total numbers of Kb-N4+ CD8+ T cells were decreased when the mice were injected with Marilyn Tregs (fig. S3, E and F). Moreover, the presence of Marilyn Tregs reduced the number of low multimer–binding T cells, as compared with Marilyn TH cells, which resulted in an increase in the relative affinity of the response (fig. S3, E, G, and H). We concluded that the adoptive transfer of antigen-specific Tregs preferentially inhibits low-avidity as opposed to high-avidity CD8+ T cells, thereby increasing the overall avidity of the T cell response.

To directly test whether Tregs preferentially inhibit the priming of low-avidity CD8+ T cells, we used two altered peptide ligands that are recognized with different affinities by the OT-I TCR [OT-I is a CD8+, TCR-transgenic line specific for the H-2Kb-SIINFEKL (N4, OVA peptide)]. Kb-N4 complexes bind the OT-I TCR with high affinity. SIITFEKL (T4) binds Kb with an affinity similar to that for N4, but the Kb-T4 complexes are recognized by the OT-I TCR with an affinity lower than Kb-N4 complexes by a factor of 70.7 (1618). We first verified that the numbers of Kb-N4 and Kb-T4 complexes present on the DC surface were similar (fig. S4, A and B). Immunization with N4-loaded DCs (N4-DCs) induced effective OT-I expansion both in the presence and absence of Tregs, although Treg depletion increased the expansion at low peptide concentrations (Fig. 2A and fig. S4, C and D). T4-loaded DCs, as shown previously (18), induced very low expansion of OT-I cells in the presence of Tregs, even at high peptide concentrations (Fig. 2B and fig. S4, C and D). In Treg-depleted mice, by contrast, T4-DCs induced effective expansion of OT-I cells over a wide range of peptide concentrations (Fig. 2B). Similar results were obtained when we assayed OT-I activation using CD69 expression (fig. S5A), which suggests that Tregs inhibit some early steps of T cell priming by low-affinity peptides.

Fig. 2

Treg suppression inhibits T cell responses to low-affinity ligands by destabilizing T cell–DC interactions. (A and B) DEREG (black) or control (white) littermate mice were DT-injected and immunized by foot pad (f.p.) injection with mature DCs loaded with the indicated concentrations of the native N4 or the alternate peptide ligand T4. Eighteen hours later, mice were injected intravenously (i.v.) with 105 naïve CD45.1 OT-I cells. OT-I cell numbers from the harvested dpLNs 5 days after the immunization are reported. (C to H) TPLSM analysis of OT-I cell priming in DT-treated littermate and DEREG mice. DsRed+ DCs loaded or not with 1 μM of N4 or T4 peptides and unpulsed CFP+ DCs were coinjected in the f.p. of DT-treated littermate or DEREG mice in the presence of monoclonal antibodies (mAbs) that block CCL-3/4/5 or the isotype-matched control antibodies. Twenty hours later, GFP+ OT-I cells were adoptively transferred. Before in vivo imaging, mice received the l-selectin–specific antibody to block lymphocyte homing, and dpLNs were imaged 2 to 6 hours after OT-I cell transfer. (C) Representative TPLSM images are shown. White lines represent migratory paths of OT-I cells. (D) T cell mean velocities are represented; the red line indicates the median value. (E) Average contact duration. (F to H) The percentage of remaining conjugates for interactions between GFP+ OT-I cells and DsRed+ DCs are shown. Data are representative of at least three independent experiments. Error bars represent means ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; n.s., not significant.

To test if Tregs suppress priming by low-affinity peptide by acting directly on CD8+ T cells, we next reconstituted the inhibition in vitro. Tregs, but not CD4+ TH cells, inhibited the activation of OT-I cells by N4-DCs only at low peptide concentration (fig. S5B). In the case of T4-loaded DCs, the inhibition by Tregs was more potent and was observed at both high and low peptide concentrations (fig. S5C). These results suggest that the expansion of low-avidity polyclonal T cells observed in DEREG mice is due to the release of the suppression by Tregs of low-avidity T cell clones.

Tregs were previously shown to regulate DC–T cell interactions during priming (3, 5) and to suppress the production by DCs of CCL-3/4 (19, 20), a family of chemokines involved in the control of the dynamics of T cell priming (2123), as well as in the stability and signaling at DC–T cell synapse (24, 25). We therefore hypothesized that Tregs destabilize low-avidity DC–T cell interactions more efficiently than high-avidity interactions through the regulation of chemokine production.

To investigate this possibility, we first analyzed the production of CCL-3/4/5 during T cell priming in vivo by DCs loaded with N4 or T4 peptides, in the presence or absence of Tregs (fig. S6, A to C). In littermates, unloaded LPS-treated DCs induced the production of all three chemokines in draining popliteal lymph nodes (dpLNs). Loading of the DCs with T4 hardly modified chemokine production, whereas loading with N4 increased the production of CCL-3/4 (fig. S6, A and B). Treg depletion did not modify the expression of activation markers by the adoptively transferred DCs (fig. S5, D to H). By contrast, it increased CCL-3/4 production in mice injected with unloaded or T4-loaded DCs but not with N4-DCs (most likely because the production of these chemokines was already very high).

We therefore investigated the effects of Treg depletion and the eventual role of these chemokines on DC–T cell interactions induced by high- or low-affinity peptides using intravital dynamic two-photon laser scanning microscopy (TPLSM). Littermates or DEREG mice treated with DT were adoptively transferred with a 1:1 mix of DsRed fluorescent protein (DsRed)–-expressing DCs pulsed with either N4 or T4 and unloaded control cyan fluorescent protein–positive (CFP+) DCs (as a control T cell–DC contacts). GFP+ OT-I T cells were then adoptively transferred to the mice. The possible involvement of CCL-3/4/5 was addressed by using a previously described mix of blocking antibodies (21, 23).

The injection of blocking antibodies had no major effect on DC and OT-I recruitment to the dpLNs, as compared with the effects on control mice (fig. S7, A and B). In these mice, N4-DCs caused a marked arrest of OT-I cells and long-lasting DC–T cell contacts, as compared with T4-DCs (movies S2 and S4 and Fig. 2, C to H), which confirmed that high-affinity peptides induce longer-lasting DC–T cell contacts with naïve T cells than do low-affinity peptides (16, 26). Depletion of Tregs did not affect the arrests and long-lasting DC–T cell contacts observed with N4-DCs (fig. S7C). By contrast, the intermediate OT-I mean velocity (Vmean) observed in mice injected with T4-DCs was reduced upon depletion of Tregs (Fig. 2, C and D, G, and H; movies S4 and S6), whereas the average contact duration of the individual contacts was increased (Fig. 2, E to H) (27). Therefore, the low-affinity T4 peptide mediates relatively stable DC–OT-I interaction only in the absence of Tregs, consistent with the observation that it only induces expansion after Treg depletion.

Treatment of the mice with CCL-3/4/5–blocking antibodies had no major effect on OT-I displacement or contact durations in the case of N4-DCs (movie S1 and Fig. 2, D to F). In mice injected with T4-DCs, by contrast, the blocking antibodies reversed the effects of Treg depletion, both in terms of reduction of the Vmean and of the duration of the individual contacts (Fig. 2, D to E, G, and H; movies S3 and S5). The lack of effect of the chemokine antibodies in mice injected with N4-DCs could be because of the higher levels of chemokine production observed in these mice (fig. S6, A to C), or stable N4-DC–T cell conjugates may become chemokine-independent. These results suggest that the stable interactions of T cells with DCs bearing low-affinity peptides that occur in the absence of Tregs require CCL-3/4/5. We propose that by limiting the production of CCL-3/4/5, Tregs normally inhibit stable interactions between DCs and low-avidity T cells, thereby limiting their priming.

To investigate the role of Tregs during CD8+ T cell responses to a microbe-associated nonself antigen, we infected DT-treated DEREG mice or littermates with Listeria monocytogenes (LM) expressing recombinant OVA (rLM-OVA) (28). Treg depletion affected neither the proportion of myeloid cells (fig. S8, A to F) nor the bacterial burden at days 3 and 5 postinfection, or the capacity of the mice to clear the infection (Fig. 3A) [clearance of the primary LM infection is known to be mainly mediated by innate immunity (29)]. Treg depletion caused a reduction in the total numbers of Kb-N4 multimer–positive T cells at day 7 (peak of the response, Fig. 3B). In the presence of Tregs, rLM-OVA infection induced primarily a population of high multimer–binding T cells (Fig. 3C). Upon Treg depletion, the proportion of low multimer–binding cells increased (Fig. 3, C and D, and fig. S9, A to D). These cells were antigen-specific, as they did not bind a control Kb-SSIEFARL multimer (fig. S9A). Although the total numbers of both high and low multimer–binding cells decreased slightly in Treg-depleted mice, the decrease was significantly greater for the high multimer–binding cells (Fig. 3D). This decrease could be because of overstimulation of high-avidity T cells (which may impair T cell activation (30)] or to competition of the OVA-specific cells with the very numerous T cells that respond to other LM antigens in the absence of Tregs (fig. S10, A and B).

Fig. 3

Tregs promote high-avidity CD8+ T cell primary immune responses to rLM-OVA infection. DT-treated DEREG and littermate mice were injected i.v. with 5 × 103 rLM-OVA. (A) At days 3, 5, and 7 after infection (p.i.), the number of live bacteria per spleen (left) and liver (right) was determined as CFU, colony-forming units. (B) Primary CD8+ T cell responses were measured after the infection by Kb-N4–multimer staining. The number of CD8+ Kb-N4+ cells is reported. (C) CD8+-gated flow cytometry plots from samples collected day 7 after infection are shown. (D) The numbers of high-avidity and low-avidity antigen-specific cells were determined by Kb-N4 multimer staining. (E) Relative affinity for Kb-N4+ cells was determined after early (DT days –1, 0) or late depletion (DT days 3 to 4). (F) Ex vivo multimer-dilution assay at day 7 after rLM-OVA infection. The graph shows the percentage of Kb-N4 multimer–positive on gated CD8+ T cells, normalized to the number of Kb-N4 multimer–positive cells at the highest multimer concentration. Individual mice (black line) and the mean values (red line) are represented. (G) For each individual mouse, the data obtained in (F) were analyzed to fit to sigmoid dose-response curves and the EC50 value was calculated. Tregs were depleted at days –1 and 0 (early depletion) or at days 3 and 4 postinfection (late depletion). Data are representative of three independent experiments. Error bars represent means ± SEM. N.s., not significant. *P < 0.05; **P < 0.01; n.s., not significant.

As expected from the changes in the proportions of high and low multimer–binding cells, Treg depletion induced a decrease in the relative affinity of the OVA-specific–responding T cells (Fig. 3E and fig. S9, C and D). Decreased multimer labeling was not due to lower levels of CD8 expression (fig. S9B). Decrease in the avidity of the OVA-specific CD8+ T cell response upon depletion of Tregs was also evident in the multimer dilution assay (Fig. 3, F and G). Interferon-γ (IFN-γ) production by CD8+ T cells was shown previously to correlate with the avidity of LM-specific immune responses (10). Both the number of OVA-specific IFN-γ–producing CD8+ T cells and the amount of IFN-γ per cell (MFI) were lower when the infection took place in the absence of Tregs (fig. S11, A to C). When Treg depletion was delayed to days 3 to 4 after infection, the relative affinity and the EC50 of multimer binding assay (Fig. 3, E and G), as well as the numbers of Kb-N4+ multimer+ and IFN-γ+ cells, were all unaffected, whereas the MFI of the IFN-γ labeling was slightly increased (fig. S12, A to C).

To investigate the mechanisms involved in the beneficial effects of Tregs during priming to LM antigens, we harvested the spleens of mice infected in the presence or absence of Tregs and analyzed different inflammatory mediators. No major effects were seen on the secretion of CXCL1 and CXCL10) (fig. S13, A and B). By contrast, the production of CCL-2/3/4 was increased upon Treg depletion in vivo (fig. S13, C to E). Furthermore, upon ex vivo restimulation with heat-killed (HK)-rLM-OVA, the splenocytes isolated from infected mice showed substantially enhanced secretion of CCL-3/4/5 when the infection in vivo took place in the absence of Tregs (fig. S14). These findings indicate that, as shown previously by others (19, 20), Tregs inhibit in vivo the production of CCL-2/3/4/5.

Because our previous results implicate CCL-3/4/5 in the control of the priming of low-avidity T cells, we decided to explore the possible involvement of the overproduction of these chemokines in the detrimental effects of Treg depletion on T cell priming during LM infection. To do so, we injected a mix of blocking antibodies to CCL-3/4/5 after Treg depletion and infection with rLM-OVA (21, 23). In both isotype control and blocking antibody–treated mice, the bacteria were undetectable in the spleen at day 7. By contrast, only injection of the blocking antibodies reversed the effects of Treg deprivation on the proportion of low and high multimer–binding cells and of the relative affinity of the responding OVA-specific T cells (fig. S15, A and B). These results suggest that CCL-3/4/5 production in the absence of Tregs is required for the observed reduction in avidity of the OVA-specific CD8+ T cells after LM infection.

We next investigated the functional relevance of the presence of Tregs during the primary infection by rLM-OVA. Because the avidity of primary responses was suggested to influence memory responses (12, 15), we investigated the memory responses generated after LM infection in Treg-deprived mice. Littermates and DEREG mice treated with DT were primed with rLM-OVA and tested for memory responses by reinfecting the mice with rLM-OVA 50 days later. After secondary infection, 10 to 25 times more bacterial burden was detected in both spleen and liver when Tregs had been depleted before the primary infection (Fig. 4A). Consistent with this result, both the number of memory Kb-N4 multimer–positive T cells (Fig. 4, B and C) and their relative affinity (Fig. 4D) were reduced. After ex vivo restimulation, we found reduced numbers of IFN-γ–granzyme B double-positive memory CD8+ cells (Fig. 4, E and F). Therefore, the presence of Tregs during the primary rLM-OVA infection is required for the establishment of fully effective high-avidity CD8+ T cell memory responses.

Fig. 4

During T cell priming, Tregs are required to generate a protective immune response after secondary rLM-OVA challenge. Littermates and DEREG mice were treated with DT and infected with rLM-OVA, and tested for memory protection 50 days later. For memory T cell generation, naïve and 50-days postinfection mice were given 5 × 103 and 2 × 105 rLM-OVA, respectively. (A) The number of live bacteria per spleen (left) and liver (right) was determined 3 days after secondary challenge. As control, C57BL/6 mice that had not received primary infection were similarly infected. In 10 of 12 spleens from littermate mice, the bacteria were undetectable. (B to F) Kb-N4+ multimer and intracellular IFN-γ and granzyme B staining were done to determine the percentage of antigen-specific cells in CD8+ T cell population. (B) CD8+-gated flow cytometry plots are shown (samples were collected 3 days after challenge). (C) The number of Kb-N4+ cells is shown. Individual mice are represented, control uninfected (gray), littermate (white), and DEREG (black). Results from three independent experiments are shown. (D) Relative affinity for Kb-N4+ cells was determined. (E and F) After ex vivo N4 restimulation, IFN-γ and granzyme B staining was done to determine the number of CD8+ antigen-specific cells. (E) CD8+-gated flow cytometry plots are shown. (F) The number of IFN-γ+ granzyme B+ CD8+ cells is shown. Data are representative of three independent experiments. Error bars represent means ± SEM. **P < 0.01; ***P < 0.001.

We propose that the absence of Tregs reduces the “fitness” of primary CD8+ T cell responses, causing the overproliferation of low-avidity T cells, and may impair the activation of high-avidity T cells, although this remains to be explored. The inhibition of low-avidity T cell clones by Tregs could also help explain why Tregs fully control T cell reactivity to self-antigens [which are generally of low avidity because of negative selection (10)], while sparing T cell responses to nonself antigens (which target a nonnegatively selected repertoire that includes high-avidity T cells). These results unravel an unexpected function for Tregs during CD8+ T cell priming and should inform Treg manipulation for the design of long-term vaccination.

Supplementary Materials

www.sciencemag.org/cgi/content/full/338/6106/532/DC1

Materials and Methods

Figs. S1 to S15

References (3135)

Movies S1 to S6

References and Notes

  1. Acknowledgments: We thank all the U932 and TWINCORE Sparwasser members; D. Raulet, A. Dielmann, P. Paul-Gilloteaux, O. Lantz, and M. Albert; the Nikon Imaging Center and flow cytometry platform; and the animal facilities of Institut Curie and TWINCORE for help with the experiments and helpful discussions. rLM-OVA was kindly provided by H. Shen. This work was supported by funding from the Institut Curie; Institut National de la Santé et de la Recherche Médicale; Centre National de la Recherche Scientifique; La Ligue Contre le Cancer; Association de Recherche Contre le Cancer (ARC); EC grant ENCITE, Health-F5-2008-201842, SFB 900. L.P. has been supported by ARC, A.T. was a fellow of Ministère de l’Education et de la Recherche. C.A.-S. was supported by the Boehringer Ingelheim Fonds, Foundation for Basic Research in Medicine. The authors have no conflicting financial interests. The data are tabulated in the main paper and in the supplementary materials.
View Abstract

Navigate This Article