Rapid Functional Decline of Activated and Memory Graft-versus-Host–Reactive T Cells Encountering Host Antigens in the Absence of Inflammation

Graft-versus-host disease (GVHD) complicating allogeneic hematopoietic cell transplantation offsets its beneficial graft-versus-leukemia (GVL) effects, both of which largely reflect GVH alloresponses (1). Conditioning-induced injury (2), which results in translocation of LPS, a TLR4 agonist (3), promotes GVHD (2, 4), and interference with LPS–TLR4 interactions attenuates GVHD (4, 5). We have shown that administration of delayed donor lymphocyte infusions (DLIs) containing large numbers of host-reactive naive donor T cells to established mixed hematopoietic chimeras, whose conditioning-induced inflammation has subsided, does not induce GVHD, yet results in potent GVH alloresponses that convert the mixed chimeras to full donor chimerism and eliminate recipient lymphohematopoietic tumors (6–8). Due to the absence of systemic or local tissue inflammation, the GVH alloresponse is confined within the lymphohematopoietic system. The absence of GVHD in mixed chimeras receiving delayed DLI reflects failure of GVH-reactive T cells to traffic into epithelial target tissues (9). Local inflammatory stimuli promote trafficking to the site of inflammation, whereas systemic TLR stimuli promote migration into multiple GVHD target tissues (9).

Although the lack of inflammation reduces GVHD in established mixed chimeras receiving DLI, it also limits GVL effects, as GVL was less potent and more dependent on CD4 T cell help than GVL effects in freshly irradiated recipients (10). Thus, the quiescent host environment seems to attenuate the GVH alloresponse, either at the priming phase, the effector phase, or both. Indeed, impaired acquisition of effector functions by naive donor T cells was observed in mixed chimeras receiving DLI and these were restored by coadministration of a TLR agonist (10). However, GVH-reactive T cells activated in mixed chimeras receiving DLI mediated lethal GVHD when transferred to irradiated secondary hosts, demonstrating the critical impact of the host environment and the plasticity of graft-versus-host primed T cells (9). Consistently, transfer of GVH-reactive primed T cells recovered from freshly irradiated hosts developing GVHD into mixed chimeras did not induce GVHD (9). This observation raises the question of how the quiescent host environment influences an established effector GVH alloresponse.

We have now investigated the effects of the quiescent environment on the effector functions of GVH-reactive CD4 and CD8 primed T cells. We generated GVH-reactive primed T cells by administering naive allogeneic donor T cells to lethally irradiated bone marrow transplant recipients, then isolated donor GVH-reactive primed T cells 4 d posttransfer from the recipients’ spleens (9). These primed T cells were transferred to mixed chimeras and chimerism and primed T cell fates were analyzed. We found that the quiescent host environment led to loss of some effector functions of GVH-reactive primed T cells and impeded the generation of effective memory. These results have important implications for T cell–mediated tumor immunotherapy.

Female B6D2F1 (BDF1; H-2^b/d), C57BL/6 (B6; H-2^b, CD45.2⁺), IFN-γ knockout B6, CD45.1⁺ congenic, and GFP transgenic B6 mice, aged 6–12 wk, were purchased from the Jackson Laboratory. Protocols involving the use of animals were approved by the Massachusetts General Hospital and Columbia University Medical Center Subcommittees on Research Animal Care, and all of the experiments were performed in accordance with the protocols.

To generate mixed chimeras, we gave BDF1 mice 150 μg PK136 Ab i.p. on day −1 to deplete NK cells; then mice received 3 Gy total body irradiation using a [¹³⁷Cs] irradiator or an X-ray irradiator (RadSource Technologies) followed by injection of 1.5 × 10⁷ T cell–depleted (TCD) bone marrow cells (BMCs) from B6 mice i.v. on day 0. Bone marrow T cell depletion was performed using anti-CD4 and anti-CD8 microbeads (Miltenyi Biotec). Chimerism was assayed by flow cytometry at various time points posttransplantation. Established mixed chimeras were used 6 wk after bone marrow transplantation for adoptive transfer experiments.

BDF1 mice received 10.25 Gy total body irradiation from a [¹³⁷Cs] or X-ray irradiator (Rad Source Technologies) followed by injection of 6–10 × 10⁶ purified B6 CD45.1⁺ splenic T cells (equivalent of 3 × 10⁷ splenocytes) and 5 × 10⁶ TCD BMCs from wild-type CD45.2⁺ B6 mice. T cells were enriched using the untouched T cell isolation kit, with purity >95%. Four days later, these mice were sacrificed and splenic T cells were isolated using the untouched T cell isolation kit (Miltenyi Biotec). Residual BDF1 cells were depleted using a biotin-conjugated Ab specific for H-2D^d (34-2-12) and anti-biotin microbeads (Miltenyi Biotec). The purified primed T cells contained ∼54% CD4 and 44% CD8 donor-derived T cells. A total of 2 × 10⁶ purified GVH-reactive primed T cells were transferred to lethally irradiated (10.25 Gy) secondary BDF1 mice or established mixed chimeras i.v. TCD BMCs from CD45.2⁺ B6 mice were also given to the secondary BDF1 recipients with the T cells to avoid marrow aplasia. In some experiments, CD4⁺ or CD8⁺ GVH-reactive primed T cells were further enriched by depletion of CD8 or CD4 cells using anti-CD8 or anti-CD4 microbeads (Miltenyi Biotec).

In vivo cytotoxicity assay was performed as reported previously (10). In brief, B6 and BDF1 splenocytes labeled with 0.5 and 5 μM CFSE (Invitrogen), respectively, were then mixed at a ratio of 1:1 and injected i.v. (total 2 × 10⁷/recipient). Eighteen hours later, recipient mice were sacrificed and harvested splenocytes were analyzed by flow cytometry for the presence of the two populations. The index of survival was defined as the percentage of allogeneic B6DF1 splenocytes (CFSE^high) divided by the percentage of B6 splenocytes (CFSE^low) as described previously (11).

GVH-reactive primed T cells were labeled with 5 μM CFSE (Invitrogen) and injected i.v. into secondary recipient mice. Proliferation of transferred CD4 and CD8 T cells in various lymphoid tissues as determined by CFSE dilution was determined by flow cytometry at various time points posttransfer.

Donor chimerism in peripheral blood was determined by flow cytometry as reported before (9). Annexin V-allophycocyanin was used to detect early apoptosis in DAPI⁻ cells. Labeled cells were analyzed with a FACSCanto II flow cytometer (BD Biosciences). Transferred primed T cells were excluded by an anti-CD45.1 Ab, and donor chimerism was determined in CD45.1⁻ cells. To detect IFN-γ production, we stimulated splenocytes either with 50 ng/ml PMA (Sigma-Aldrich) and 500 ng/ml Ionomycin (Sigma-Aldrich) or with BDF1 splenocytes in vitro for 5 h in the presence of brefeldin A (Biolegend). In some studies, splenocytes were stimulated with BDF1 splenocytes for 24–48 h and brefeldin A was added for the last 5 h. Intracellular staining was then performed to detect production of IFN-γ by CD4 and CD8 CD45.1⁺ primed T cells.

LPS (Escherichia coli 055:B5) was purchased from Sigma-Aldrich and was injected i.p. at the dose of 50 μg/mouse in 0.5 ml PBS on days 0, 2, and 4 after primed T cell transfer.

Data were analyzed with Prism 4.0 statistical software. Group comparisons were made by two-tailed Student t test or one-way ANOVA (Dunnett multiple comparison test). A p value < 0.05 was considered statistically significant. Data are presented as mean ± SEM.

We previously showed that transfer of GVH-reactive primed T cells recovered from mice developing GVHD failed to induce GVHD when transferred to mixed chimeras, whereas these cells caused severe GVHD when transferred to freshly irradiated secondary recipients (9). To determine whether primed T cells still mediated lymphohematopoietic GVH alloresponses after transfer, we assessed the increase in donor chimerism in mixed chimeric recipients of transferred cells. To generate GVH-reactive primed T cells, lethally irradiated BDF1 recipients received 6–10 × 10⁶ CD45.1⁺ congenic B6 T cells (equivalent of 3 × 10⁷ splenocytes) along with TCD B6 BMCs i.v. Four days later, in vivo cytotoxicity studies were performed or animals were sacrificed, splenocytes were harvested, and GVH-reactive primed CD45.1⁺ B6 T cells were isolated. Potent in vivo cytotoxicity against recipient cells was observed (Fig. 1A), and phenotypic studies revealed expression of the activation marker CD25 (Fig. 1B) and a mainly CD44^highCD62L^low effector memory phenotype of GVH-reactive CD45.1⁺ B6 T cells (Fig. 1C). Compared with naive T cells, these primed T cells showed increased IFN-γ production (Fig. 1D).

To determine whether or not these GVH-reactive primed T cells can increase donor chimerism in quiescent mixed chimeric recipients, we generated mixed chimeras by transplantation of B6 BMCs to BDF1 mice receiving nonmyeloablative conditioning 6 wk earlier. After 6 wk, primed T cells (2 × 10⁶/mouse) were transferred to these established mixed chimeras or freshly irradiated BDF1 hosts (Fig. 2A).

As shown in Fig. 2B, naive DLI led to rapid conversion of mixed chimeras to full donor chimerism, as we have previously reported (6–8, 12). Donor chimerism in B220 and CD3 lineages was also increased in mixed chimeras receiving primed T cells, but to a lesser degree (p < 0.05, by day 14 and later), consistent with our previous findings (9) and despite the potent anti-BDF1 cytotoxicity (Fig. 1A) and the greater activation of GVH-reactive T cells in primed T compared with naive T cell inocula (Fig. 1B, 1C). Donor chimerism in Mac-1⁺ cells increased to 95% in all (4/4) recipients of naive T cells, but only in 1 of 7 recipients of primed T cells (Fig. 2B). Similar results were obtained in a replicate experiment (Supplemental Fig. 1). Thus, the decreased ability of primed T cells to markedly increase donor chimerism in established mixed chimeras suggested that their effector function may be impaired by transfer to mixed chimeras.

To determine the impact of the host environment, we compared the phenotypes of primed T cells after transfer to established mixed chimeras and freshly irradiated recipients. CD25 expression increased further on both CD4 and CD8 primed T cells by 4 d after transfer to freshly irradiated recipients, then declined by day 10. In contrast, expression of CD25 on CD4 and CD8 primed T cells transferred to quiescent mixed chimeras decreased markedly and was significantly lower than that of primed T cells in freshly irradiated recipients at both time points (p < 0.05; Fig. 3A, 3B).

After transfer to freshly irradiated recipients, the percentage of CD44^highCD62L^low CD4 and CD8 effector cells increased, whereas the percentage of CD44^highCD62L^high central memory primed T cells decreased. Interestingly, phenotypic changes of GVH-reactive primed CD4 T cells transferred to mixed chimeras showed similar kinetics to those in freshly irradiated recipients, although with a slightly lower percentage of CD44^highCD62L^low effector T cells and higher percentage of CD44^highCD62L^high central memory T (Tm) cells in mixed chimeras on day 4 (Fig. 3C, 3D). In contrast, primed CD8 T cells showed markedly increased proportions of central memory and decreased proportions of effector memory cells after transfer to quiescent mixed chimeras compared with freshly irradiated hosts (Fig. 3C, 3D). Thus, transfer of primed T cells to the quiescent environment in mixed chimeras led to an increase in central/memory and a decrease in effector phenotype compared with transfer to freshly irradiated hosts.

We compared the proliferation and accumulation of CFSE-labeled primed T cells posttransfer to freshly irradiated recipients and established mixed chimeras. CD4 and CD8 primed T cells in spleen, bone marrow, and lymph nodes of freshly irradiated adoptive recipients had undergone extensive division and were CFSE⁻ by day 4. In contrast, significantly less cell division had occurred among primed CD4 and CD8 T cells transferred to established mixed chimeras by day 4 (p < 0.05; Fig. 4A, 4B). By day 10, most CD4 and CD8 primed T cells in mixed chimeras had also undergone extensive proliferation, although the percentages of CFSE⁻ CD4 primed T cells in spleen and bone marrow remained slightly lower than those in irradiated recipients. The recovery of transferred primed T cells in spleen, bone marrow, lymph nodes, and peripheral blood was lower in mixed chimeras on day 4 (p < 0.05), but was either greater than or similar to that in freshly irradiated adoptive recipients by day 10 (Fig. 4C). Thus, expansion of primed T cells in quiescent mixed chimeras was delayed but not decreased compared with that in freshly irradiated adoptive recipients. Analysis of apoptosis of primed T cells at 30 min (Supplemental Fig. 2) or on days 4 and 10 (Fig. 4D) after transfer did not show differences between freshly irradiated mice and mixed chimeras.

The conversion to full donor chimerism in mixed chimeras receiving GVH-reactive naive or primed T cells depends on the destruction of host hematopoietic cells by the GVH-reactive T cells. Because there was no increase in apoptosis or failure of proliferation of primed T cells transferred to established mixed chimeras compared with freshly irradiated recipients, we considered the possibility that GVH-reactive primed T cells lost their effector function after transfer to mixed chimeras. As shown in Fig. 5A and 5B, primed T cells transferred to freshly irradiated recipients demonstrated potent cytotoxicity posttransfer, as indicated by the low BDF1 cell survival compared with that in freshly irradiated control mice not receiving primed T cells (p < 0.0001). Because the GVH-reactive primed T cells recovered from the primary recipients had potent cytotoxicity before transfer (Fig. 1A), the results indicate that the effector function of these primed T cells was maintained after transfer to irradiated recipients. In contrast, in vivo cytotoxicity was lost when primed T cells were transferred to mixed chimeras, because BDF1 cell survival in mixed chimeras receiving primed T cells was not significantly different from that in mixed chimeras not receiving primed T cells and was significantly greater than that in irradiated recipients of primed T cells (p < 0.001 on day 3 and p < 0.00001 on day 9; Fig. 5B). These data indicate that antihost cytotoxic activity of primed T cells was lost when they were transferred to mixed chimeras, but not when they were transferred to irradiated recipients.

Cytokine production was determined on days 4 and 10 posttransfer of primed T cells. As shown in Fig. 5C and 5D, production of IFN-γ by primed CD4 and CD8 T cells transferred to mixed chimeras was not different from that of primed T cells transferred to freshly irradiated recipients. Thus, primed T cells lost their cytotoxic effector function, but not their cytokine-producing activity after transfer to the quiescent environment of established mixed chimeras.

We hypothesized that a lack of irradiation-induced inflammation in mixed chimeras might cause the loss of effector function observed in transferred GVH-reactive primed T cells. To address this hypothesis, we injected bacterial LPS, a TLR4 agonist (13) that promotes GVHD (2, 4, 5), to mixed chimeras on days 0, 2, and 4 after transfer of GVH-reactive primed T cells to provide a systemic inflammatory stimulus. Whereas donor chimerism was modestly increased in mixed chimeras receiving primed T cells alone, a marked, rapid increase in donor chimerism was observed in mixed chimeras receiving primed T cells with LPS (p < 0.05), with conversion to full donor chimerism by day 14, which was more rapid than the conversion induced by naive T cells (Fig. 6). Mixed chimeras receiving LPS alone did not show an increase in donor chimerism. Thus, the increase in donor chimerism required administration of both primed T cells and LPS. These data suggest that provision of inflammatory stimuli to mixed chimeras preserves cytotoxic effector function of primed T cells after adoptive transfer.

The ability of transferred primed CD8 T cells to produce cytokines (Fig. 5), even in the absence of TLR stimuli, seemed to conflict with previous findings indicating that inflammation was needed for IFN-γ production by primed CD8 T cells (14, 15). One difference distinguishing our studies from others (14, 15) is that primed CD4 T cells were cotransferred in our studies. We therefore investigated whether the cotransfer of GVH-reactive primed CD4 T cells with primed CD8 T cells was required for the intact cytokine production of primed CD8 T cells after transfer to quiescent mixed chimeras. We transferred primed CD4, CD8, or total (CD4 and CD8) T cells to established mixed chimeras and analyzed the phenotype, accumulation, and ability of primed T cells to produce IFN-γ 4 and 10 d later. Expression of CD25 (Fig. 7A) and the effector/memory phenotype (Fig. 7B, 7C) of primed CD4 or CD8 T cells transferred alone was similar to that after transfer of total primed T cells, although primed CD8 T cells transferred alone had a higher percentage of central Tm cells on day 10 compared with recipients of both subsets (p < 0.05; Fig. 7C). However, in the absence of primed CD4 T cells, the accumulation of primed CD8 T cells was seriously impaired, with markedly reduced numbers in tissues of mixed chimeras receiving primed CD8 T cells alone compared with mice receiving both subsets (p < 0.05; Fig. 7D). Interestingly, the accumulation of primed CD4 T cells was decreased in the absence of primed CD8 T cells, both in the spleen on day 4 and in blood on day 10 (p < 0.05; Fig. 7D). Importantly, in the absence of primed CD4 T cells, primed CD8 T cells showed impaired IFN-γ production in response to recipient Ag stimulation at both time points (p < 0.05; Fig. 7E). However, no such defect was found when primed CD8 T cells were stimulated with PMA and Ionomycin, suggesting that the impaired IFN-γ production was not due to loss of effector function of the primed CD8 T cells, but rather to loss of Ag responsiveness (Fig. 7F). IFN-γ production by primed CD4 T cells in response to recipient alloantigen stimulation was also impaired on day 10 in the absence of primed CD8 T cells (p < 0.05; Fig. 7E), and this was also reversed by stimulation with PMA and Ionomycin (Fig. 7F). These data not only demonstrated that help from primed CD4 T cells enabled primed CD8 T cells to retain their ability to accumulate and produce IFN-γ in response to recipient Ags in the absence of overt inflammation, but also indicated that primed CD8 T cells could exert functionally enhancing effects on primed CD4 T cells, although to a lesser degree. Thus, optimal expansion and cytokine production of GVH-reactive primed CD4 and CD8 T cells in response to recipient alloantigens after transfer to mixed chimeras required interactions between the two T cell subsets.

We further investigated the mechanisms of the help provided by primed CD8 T cells to primed CD4 T cells. Because CD8 T cells are the major source of alloantigen-induced IFN-γ (10, 16) which has been shown to be involved in modulation of responses of CD4 T cells by CD8 T cells in other settings (17–19), we hypothesized that IFN-γ derived from primed CD8 T cells mediated help to primed CD4 T cells after transfer to established mixed chimeras. To address this hypothesis, we generated GVH-reactive primed T cells with CD45.1⁺ congenic B6 and IFN-γ knockout B6 T cells. CD45.1⁺ primed CD4 T cells were then injected alone, or together with CD45.1⁺ wild type or IFN-γ knockout primed CD8 T cells to established mixed chimeras. On days 4 and 10 posttransfer, accumulation of CD45.1⁺ primed CD4 T cells in lymphoid tissues and IFN-γ production in response to alloantigen stimulation were determined. As shown in Fig. 7G, accumulation of primed CD4 T cells in spleen on days 4 and 10 posttransfer was significantly higher in mixed chimeras receiving primed CD4 and CD45.1⁺ CD8 T cells than in those receiving primed CD4 T cells alone. Cotransfer of IFN-γ primed CD8 T cells did not significantly increase accumulation of primed CD4 T cells compared with recipients of primed CD4 T cells alone. In addition, although IFN-γ production by primed CD4 T cells transferred alone to established mixed chimeras was significantly decreased compared with primed CD4 T cells transferred with CD45.1⁺ primed CD8 T cells, cotransfer of IFN-γ knockout primed CD8 T cells did not restore IFN-γ production by primed CD4 T cells on day 10 posttransfer (Fig. 7H). Thus, our results showed that primed CD8 T cell–derived IFN-γ mediated T cell help to primed CD4 T cells and promoted CD4 effector function after transfer to established mixed chimeras.

Ag-specific responses generate long-lived Tm cells (20, 21). Although TLR stimulation is known to enhance the generation of memory from naive T cells (22–24), the influence of TLR stimulation during the effector phase on the generation of functional Tm is unknown. To address this issue, we analyzed the accumulation, function, and phenotype of GVH-reactive Tm cells in the experiment presented in Fig. 6. Four months after transfer of GVH-reactive CD45.1⁺ primed T cells with or without LPS, we enumerated CD45.1⁺ T cells, which were all CD44⁺ Tm cells at this point, in spleen, lymph nodes, peripheral blood, and bone marrow of mixed chimeras. Mixed chimeras receiving primed T cells and LPS all converted to full donor chimerism, whereas mixed chimeras receiving primed T cells alone remained mixed chimeric in multiple lineages, with donor chimerism ranging from 5 to 66% (data not shown). We detected similar numbers of CD45.1⁺ Tm cells in all of these tissues of both groups, except that the number of CD8 Tm cells in bone marrow was significantly higher in mixed chimeras that had received LPS at the time of original primed T cell transfer (Fig. 8A). The majority of CD4 and CD8 Tm cells showed a CD44⁺CD62L⁻ effector memory phenotype in both groups (Fig. 8B). Expression of PD-1 on CD4 and CD8 Tm cells from mixed chimeras receiving or not receiving LPS was significantly greater than that on naive CD4 and CD8 T cells. However, PD-1 expression was significantly lower on CD4⁺ Tm cells from mixed chimeras that had received primed T cells with LPS compared with those in mice that had not received LPS. There was no difference between the two groups in PD1 expression on CD8 Tm cells (Fig. 8C). Despite the similar accumulation and phenotype, both CD4 and CD8 Tm cells from mixed chimeras that had originally received LPS with the primed T cell transfer demonstrated significantly greater expression of CD25 and IFN-γ production than those from mixed chimeras that had received these cells without LPS, when stimulated with BDF1 recipient Ag in vitro (Fig. 8D, 8E). Thus, exposure of primed T cells to TLR stimulation after transfer to mixed chimeras not only maintained their cytotoxic function in the short term, but also permitted the generation of GVH-reactive Tm cells with superior ability to be activated and develop secondary effector function.

We investigated the impact of the host environment on the survival, expansion, function, and long-term memory function of differentiated GVH-reactive CD4 and CD8 primed T cells. Transfer of GVH-reactive primed T cells generated in lethally irradiated recipients to quiescent mixed chimeras or secondary irradiated hosts allowed us to determine the impact of inflammation on host alloantigen-primed GVH-reactive T cells in the effector phase of GVH alloresponses. Whereas previous studies (14, 15) investigated the effects of inflammation on transferred T cells primed by in vitro antigenic stimulation, GVH-reactive primed T cells in our study were generated in vivo in primary recipients with irradiation-induced inflammation and GVHD. Transfer of GVH-reactive primed T cells from recipients developing GVHD to quiescent mixed chimeras was associated with decreased ability to increase donor chimerism compared with transfer of naive T cells. Consistent with their decreased ability to convert mixed to full chimerism, primed T cells given to mixed chimeras showed impaired cytotoxicity compared with the same primed T cells transferred to freshly irradiated hosts. In contrast, cytokine production was not impaired when primed T cells were transferred to mixed chimeras compared with freshly irradiated recipients. Although proliferation of transferred primed T cells was delayed in mixed chimeras versus freshly irradiated adoptive recipients, overall proliferation was similar in the two groups and greater accumulation of primed CD8 T cells in lymphoid tissues was actually observed in mixed chimeras compared with freshly irradiated hosts on day 10 posttransfer. No difference was observed in apoptosis between the two groups.

The impaired cytotoxic function associated with transfer of GVH-reactive primed T cells to mixed chimeras was reversed by coadministration of the TLR4 agonist LPS to provide a systemic inflammatory stimulus, demonstrating that a proinflammatory host environment is required to maintain cytotoxic effector function of GVH-reactive primed T cells. Paulos et al. (14) and Shaulov et al. (15) previously showed that inflammation played an important role in potentiating or maintaining the in vivo expansion and cytokine production ability of Ag-specific CD8 T cells activated in vitro by their cognate Ags. In contrast with those studies, we found that GVH-reactive primed CD8 T cells had intact ability to expand and produce IFN-γ when cotransferred with GVH-reactive primed CD4 T cells. Thus, our data demonstrate that the ability of primed CD8 T cells to expand and produce IFN-γ and their cytotoxic function are differentially regulated. Although CD4 T cell help can compensate for the absence of inflammation to promote expansion and cytokine production by primed CD8 T cells, it is unable to sustain the cytotoxicity of CD8 T cells in the absence of inflammation. Although it has been well documented that TLR stimulation can potentiate the overall T cell response, the contribution of inflammation to the priming versus effector phases has not been well defined. Thus, our study demonstrates that a proinflammatory environment is required to maintain the cytotoxic function of primed CD8 T cells during the effector phase of GVHD and defines the role that CD4 T cells can play in promoting CD8 T cell function in the absence of inflammation. Although several studies have investigated the impact of CD4 T cells on the primary or memory responses of CD8 T cells (25–29), we have studied the effects of CD4 T cell help during the effector phase.

Our studies also revealed that primed CD8 T cells can promote the ability of primed CD4 T cells to accumulate and produce IFN-γ. In contrast with the well-established mechanisms of CD4 T cell help for CD8 T cells (30), mechanisms by which CD8 T cells provide help for CD4 T cells are not well defined. However, a few studies have shown that CD8 T cells are able to provide help to CD4 T cells in settings other than an alloresponse, mainly by provision of IFN-γ (17–19) and TNF-α (18). Our previous results showing that CD8 T cells are major producers of IFN-γ in mixed chimeras with graft-versus-host reactions (10) led us to hypothesize that IFN-γ might mediate CD8 help to CD4 T cells. Our results support this hypothesis, as the absence of primed CD8 T cell–derived IFN-γ led to decreased production of IFN-γ and accumulation of primed CD4 T cells, consistent with other studies (17–19). Overall, our data indicate that effector responses mediated by primed CD4 and CD8 T cells can potentiate each other and that optimal responses require both subsets.

Another new finding in our study is that the presence of inflammation during the effector phase facilitates the generation of functionally superior CD4 and CD8 Tm cells. Previous studies of the effects of TLR-induced inflammation on T cell responses did not distinguish between the priming and effector phases (22–24). Shaulov et al. (15) demonstrated that exposure of effector T cells to inflammation produced functionally superior CD8 Tm cells but did not investigate the effects of inflammation on CD4 Tm cells. To our knowledge, our data show for the first time that exposure of primed T cells to TLR4-induced inflammation led to the generation of both CD4 and CD8 Tm cells with superior functions. Moreover, the effects of inflammation on CD25 and PD-1 expression and IFN-γ production on transferred primed CD4 T cells were greater than those on CD8 Tm cells (Fig. 8C–E). The increased expression of PD-1 on both CD4 and CD8 Tm cells in mixed chimeras that did not receive LPS injection at the time of transfer suggests a similar fate as we previously observed for naive GVH-reactive T cells, which eventually became exhausted after transfer to quiescent mixed chimeras after loss of recipient chimerism (31). Exhaustion of donor GVH-reactive CD8 T cells is associated with the loss of ability to mediate GVT effects in DLI recipients after the mixed chimeras convert to full donor chimeras (6, 7). In our studies, GVH-reactive primed T cells in mixed chimeras receiving LPS were stimulated by nonhematopoietic host alloantigens after mixed chimeras were converted to full donor chimerism. In contrast, primed T cells in mixed chimeras not receiving LPS could be stimulated by both hematopoietic and nonhematopoietic host alloantigens, as they remained mixed chimeric. Therefore, the increased function of primed T cell–derived memory cells in LPS-treated fully chimeric mice is particularly striking and demonstrates the ability of early inflammation to prevent this exhaustion. Our data suggest that a proinflammatory environment during the effector phase of the GVH alloresponse might partially attenuate exhaustion of GVH-reactive DLI T cells that mediate GVL, resulting in more persistent GVL effects. Collectively, our data not only confirmed the importance of exposure to inflammation during the effector phase in the generation of CD8 Tm cells with optimal function, but also extend this finding to the CD4 T cell compartment.

The mechanisms by which inflammation is required to maintain the effector functions of primed T cells and to induce functionally superior Tm cells need further investigation. In irradiated hosts, conditioning-induced tissue damage directly induces inflammation (32), promoting release of endogenous TLR ligands and the entry of exogenous TLR ligands that promote production of proinflammatory cytokines implicated in the pathogenesis of GVHD, including IL-12 and TNF-α. IL-12 enhances and prolongs CD25 expression, and potentiates the effector functions of activated Ag-specific T cells stimulated with cognate Ag (33). Thus, reduced IL-12 levels in mixed chimeras may contribute to decreased expression of CD25 and decreased cytotoxic function of GVH-reactive primed T cells transferred to mixed chimeras. In addition, IL-12 can potentiate the generation of T cell memory responses (34, 35). Although LPS stimulation restored the cytotoxic functions of GVH-reactive primed T cells and enhanced their memory response, this effect likely involves APCs, because murine CD4 and CD8 T cells lack expression of TLR4 (36, 37). Further studies are warranted to address the role of various inflammatory cytokines in maintaining the cytotoxic function of primed CD8 T cells and in inducing functionally superior Tm cells.

In summary, our study demonstrates that an inflammatory host environment is required for the maintenance of cytotoxic effector functions of GVH-reactive primed T cells and for the generation of CD4 and CD8 Tm cells capable of rapidly developing effector function in response to antigenic restimulation. We believe our findings have implications for understanding the biology of Tm cells in general. Although a short antigenic stimulation is sufficient to establish the CD8 T cell differentiation program (38), our results underscore the importance of providing inflammatory cues during the effector phase of an immune response so that the cytotoxic function of the differentiated CD8 T cells and more functional Tm cells can be generated. Further studies are warranted to confirm these findings in other models. In addition, our findings have important clinical implications. First, we previously showed that inflammation plays a critical role in determining the strength of GVL effects when a low number of donor splenocytes containing naive GVH-reactive T cells is administered as DLI to mixed chimeras (10). The absence of inflammation in the host environment during priming reduced the effector differentiation and GVL effects of GVH-reactive CD8 T cells (10) and increased their dependence on CD4 help (39). Our current data further demonstrate that the quiescent host environment is not favorable for maintaining the cytotoxic effector function of T cells primed in an inflammatory environment, possibly contributing to the decreased GVL effects mediated by a low-dose DLI.

Second, we have previously demonstrated that although DLIs given to mixed chimeras mediate potent GVL effects, these effects extinguish over time so that animals are unable to resist tumor rechallenge (31). In settings where maintenance of the function of effector T cells and induction of Tm cell responses are critical, as in tumor immunotherapy and vaccination against infections and cancers, approaches to providing prolonged inflammatory signals may maximize the T cell effector functions and ensure a superior Tm cell response. Consistent with this notion, a recent study demonstrated that transfer of tumor-specific T cells genetically engineered to produce IL-12 mediated potent antitumor effects (40). Studies to further elucidate the impact of proinflammatory signals on the effector phase of the GVH alloresponse may enhance our understanding of T cell responses and provide novel strategies for T cell–based immunotherapies. In the context of DLI, these maneuvers would need to be balanced against the ability of inflammatory stimuli to promote GVHD (9, 31). Third, our data showing that GVH-reactive primed CD4 and CD8 T cells can provide mutual help for each other’s expansion and IFN-γ production support the notion that optimal adoptive T cell tumor immunotherapy may require the simultaneous transfer of both tumor-specific CD4 and CD8 T cells, as shown in a recent study (41).

We thank Susan Allen for expert assistance with the manuscript and Drs. Donna Farber and Ronjon Chakraverty for critical reading of the manuscript.

The authors have no financial conflicts of interest.