Sustained AhR activity programs memory fate of early effector CD8⁺ T cells

Significance

Molecular events that drive the differentiation of early CD8⁺ effector T cells toward a memory phenotype rather than terminal effector T cells are a long-standing issue in the field of immunology. In this study, we demonstrate that sustained AhR activity competes with HIF-1α to bind to HIF-1β, inducing a quiescent state in CD8⁺ effector T cells and subsequently induce the development of memory T cells, thereby revealing that AhR functions as a pivotal transcription factor that programs the memory fate of early CD8⁺ effector T cells.

Abstract

Identification of mechanisms that program early effector T cells to either terminal effector T (T_eff) or memory T (T_m) cells has important implications for protective immunity against infections and cancers. Here, we show that the cytosolic transcription factor aryl hydrocarbon receptor (AhR) is used by early T_eff cells to program memory fate. Upon antigen engagement, AhR is rapidly up-regulated via reactive oxygen species signaling in early CD8⁺ T_eff cells, which does not affect the effector response, but is required for memory formation. Mechanistically, activated CD8⁺ T cells up-regulate HIF-1α to compete with AhR for HIF-1β, leading to the loss of AhR activity in HIF-1α^high short-lived effector cells, but sustained in HIF-1α^low memory precursor effector cells (MPECs) with the help of autocrine IL-2. AhR then licenses CD8⁺ MPECs in a quiescent state for memory formation. These findings partially resolve the long-standing issue of how T_eff cells are regulated to differentiate into memory cells.

Naive T cells encounter antigens in draining lymph nodes, where they are primed by dendritic cells and differentiate into T_eff cells (1, 2). Following antigen clearance, effector cells enter a contraction phase and die, while a minority of T_eff cells survive and acquire a memory phenotype, conferring the host with the ability to defend against pathogen re-infection or tumorigenesis (3–5). CD8⁺ T_m cells are known to be directly generated from T_eff rather than from naive cells (6, 7). Notwithstanding this, a fundamental issue is how a small number of T_eff cells are selected from the effector pool to give rise to T_m cells. It is known that both low- and high-avidity T cells can yield T_m cells (8, 9), suggesting that T cell receptor (TCR)–peptide/major histocompatibility complex (MHC) interaction intensity might be dispensable and that other factor(s) are required for memory selection from the effector pool. Transcription factors (TFs) play essential roles in T cell memory development, and a panel of memory-related TFs has been identified, which are attributed to current advanced technologies such as single-cell sequencing (10, 11). However, some TFs are constitutively expressed in the cytosol, but exert their function in the nucleus; thus, they are not suitable for detecting functional alterations via genetic or epigenetic sequencing. Among them, AhR draws our attention, considering that 1) AhR plays multiple roles in modulating T cell differentiation and function (12–14); 2) inhibitory receptors such as PD-1 and LAG3 required for T cell memory pool are also regulated by AhR (15–18); 3) IL-2 signaling not only is required for the generation of memory precursor effector cells but also regulates AhR activation (19, 20); and 4) AhR plays a crucial role in detoxification which is important for cell longevity (21). In this study, we hypothesize that AhR regulates the switch from CD8⁺ effector T cells to memory cells.

Results

AhR Does Not Affect CD8⁺ Effector Cells but Is Required for Memory Formation.

CD8⁺ T cell immunity is required to control Listeria monocytogenes infection (22). In this study, we infected AhR knockout (AhR^−/−) and wild-type (WT) littermate mice with 1 × 10⁴ CFUs of Listeria monocytogenes expressing ovalbumin (Lm-OVA). The frequency of OVA-specific CD8⁺ T cells was equally distributed in the spleen, lymph nodes, blood, and lungs of WT and AhR^−/− mice seven days after infection (Fig. 1 A–C and SI Appendix, Fig. S1A). In addition, OVA-specific CD8⁺ T_eff cells produced similar amounts of IFN-γ, TNF-α and granzyme B following ex vivo SIINFEKL peptide stimulation (Fig. 1D). Stimulating WT and AhR^−/− splenic CD8⁺ T cells with anti-CD3/28 in vitro also resulted in consistent primary effector responses (SI Appendix, Fig. S1 B–D), suggesting that AhR does not affect CD8⁺ effector T cell responses. Surprisingly, 30 d later, rechallenging the above AhR^−/− mice with high-dose Lm-OVA resulted in impaired bacterial clearance and a much higher bacterial burden in the spleen and liver, compared to the littermate control mice (Fig. 1 E and F). Consistently, the number of OVA-specific T cells in the spleen and liver was reduced in AhR^−/− mice five days after the secondary infection, compared to the littermate control mice (SI Appendix, Fig. S1 E and F). However, this AhR deficiency seemed not to impair the recall response, as evidenced by the comparable expression of IFN-γ and TNF-α between the AhR^−/− and wild-type mice (SI Appendix, Fig. S1G). To determine the T cell intrinsic requirement of AhR, we adoptively transferred shAhR- and shNC-transduced OT-I T cells into mice and then infected with Lm-OVA (SI Appendix, Fig. S1 H and I). Memory formation was markedly reduced in the shAhR group after 30 d (Fig. 1 G and H). In particular, shAhR CD8⁺ T cells generated fewer CD62L⁺ memory cells than shNC CD8⁺ T cells did (Fig. 1 I and J). Apart from AhR knockout, a similar in vivo result was obtained from treatment with CH223191, a specific antagonist of AhR (SI Appendix, Fig. S1 J and K). In addition, we used IL-15 to differentiate OT-I cells into T_m in vitro based on a previously described protocol (23, 24). Consistently, we found that both AhR knockout and inhibition reduced the formation of CD62L⁺ T_m cells (Fig. 1K and SI Appendix, Fig. S1L). Together, these results suggest that AhR does not affect CD8⁺ effector cells but is required for memory development.

Fig. 1.

AhR does not affect CD8⁺ effector cells but is required for memory formation. (A–D) Experimental scheme, WT (wild-type) and AhR^−/− (AhR knockout) mice were infected with 1 × 10⁴ Lm-OVA (A). (B) frequencies of splenic CD8⁺ H-2 Kb-OVA⁺ T cells were analyzed 7 d after infection (n = 6 mice per group). (C) The number of CD8⁺ H-2 Kb-OVA⁺ T cells were analyzed 7 d after infection (n = 6 mice per group). (D) Splenocytes were stimulated with the SIINFEKL peptide for 3 h in the presence of brefeldin A. Intracellular levels of IFN-γ, TNF-α, and Granzyme B in CD8⁺ H-2 Kb-OVA⁺ T cells were measured (n = 6 mice per group). (E and F) Experimental scheme, WT and AhR^−/− mice were infected with 1 × 10⁴ Lm-OVA, followed by rechallenge of with 1 × 10⁶ Lm-OVA 30 d later (E). (E) Splenic and liver bacterial burden was measured from day 2 re-infected mice (n = 6 mice per group). (G) C57BL/6 mice were transferred with 1 × 10⁵ OT-I CD8⁺ T cells transduced with shNC or shAhR retroviruses, followed by infection of Lm-OVA. Frequencies of CD8⁺GFP⁺ OT-I T_m cells in the spleen, lymph node, blood, and lung were analyzed 30 d after infection (n = 6 mice per group). (H) The number of splenic CD8⁺GFP⁺ OT-I T_m cells was analyzed. (I) Frequencies of CD62L⁺ of CD8⁺ T_m cells in the spleen were analyzed. (J) CD62L⁺ T_m cells were enumerated 30 d after infection. (K) CD62L expression was analyzed in IL-15 differentiated WT or AhR^−/− T_m cells (n = 3). Data are representative of ≥2 independent experiments (mean ± SD). P values were calculated by two-tailed unpaired Student’s t test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; NS, no significant.

TCR Signaling-Activated Reactive Oxygen Species (ROS) Upregulates AhR in Early CD8⁺ T_eff Cells.

Next, we analyzed AhR expression in naive CD8⁺ T cells (T_n; CD62L⁺CD44⁻) and T_eff cells to explore how AhR affects memory formation. We found that AhR was barely expressed in CD8⁺ T_n cells, but was rapidly up-regulated upon anti-CD3/CD28 stimulation within 8 h (Fig. 2 A and B), as well as the expression of Cyp1a1 and Cyp1b1, two typical AhR-targeted genes (Fig. 2C). In line with the in vitro results, OT-I T cells up-regulated Cyp1a1 and Cyp1b1 in mice 24 h post Lm-OVA challenge (Fig. 2D), suggesting that AhR is rapidly induced in early CD8⁺ T_eff cells. Our previous studies reported that AhR transactivates PD-1 expression (15, 19). Coincidently, CD8⁺ T cells showed early up-regulation of PD-1 expression upon anti-CD3/CD28 stimulation (SI Appendix, Fig. S2A). In addition, AhR knockout or inhibition abrogated the up-regulation of PD-1 in early stimulated CD8⁺ T cells (SI Appendix, Fig. S2 B and C), further supporting that AhR is rapidly up-regulated in CD8⁺ T_eff cells. It has been reported that type 1 regulatory cell mobilizes STAT3 to transactivate AhR expression (12). However, we found that use of the STAT3 inhibitor STAT3-IN-1 to treat the stimulated cells did not affect AhR expression (SI Appendix, Fig. S2D). In addition to STAT3, nuclear factor-erythroid 2-related-2 (Nrf2) can also transactivate AhR expression (25, 26). We found that although hardly detectable in naive T cells, Nrf2 was rapidly up-regulated in T_eff cells (Fig. 2E and SI Appendix, Fig. S2E). Moreover, AhR expression and activity were abrogated upon Nrf2 inhibition (Fig. 2 F and G). By performing ChIP-qPCR, we also found that Nrf2 bound to the AhR prompter region (SI Appendix, Fig. S2F). Nrf2 is normally sequestered by cytoplasmic Kelch-like ECH-associated protein 1 (Keap1) and targeted to proteasomal degradation; however, ROS stress, such as H₂O₂, causes Nrf2 dissociation from Keap1 and translocation into the nucleus for transcriptional function exertion (27–29). Upon TCR engagement and T cell activation, rapid increases in ROS levels have been reported (30, 31). In line with this, we found that ROS levels were rapidly elevated after anti-CD3 and CD28 stimulation (Fig. 2H). Importantly, ROS clearance by N-acetyl cysteine (NAC) resulted in the disruption of Nrf2 nuclear localization (SI Appendix, Fig. S2G), concomitant with down-regulation of AhR expression in T_eff cells (Fig. 2 I and J). In addition, H₂O₂ treatment resulted in naive CD8⁺ T cells to up-regulate AhR expression and activation (SI Appendix, Fig. S2 H and I), which, however, was abrogated by Nrf2 inhibition (SI Appendix, Fig. S2 J and K). Together, these results suggest that AhR is rapidly up-regulated in early CD8⁺ T_eff cells through a ROS-Nrf2 mediated pathway.

Fig. 2.

TCR signaling-activated ROS up-regulates AhR in early CD8⁺ T_eff cells. (A–C) Naive CD8⁺ T cells were stimulated with anti-CD3, anti-CD28, and IL-2 (n = 3). (A and B) The expressions of AhR were analyzed by western blot (A) and qPCR (B). (C) mRNA levels of Cyp1a1 and Cyp1b1 were analyzed. (D) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA. Splenic CD45.1⁺ CD8⁺ OT-I T cells were isolated from infected mice. The expressions of Cyp1a1 and Cyp1b1 were measured (n = 3). (E) Naive CD8⁺ T cells were treated as (A), and the Nrf2 level was analyzed by western blot (n = 3). (F and G) the expression (F) and activation (G) of AhR were analyzed in ML385-treated CD8⁺ T cells (n = 3). (H) Naive CD8⁺ T cells were treated as (A), and the cellular ROS levels were analyzed by flow cytometry (n = 3). (I and J) The expression (I) and activation (J) of AhR were analyzed in NAC-treated CD8⁺ T cells (n = 3). Data are representative of ≥3 independent experiments (mean ± SD). P values were calculated using one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Activated AhR Constitutively Exists in Memory Precursor Effector Cells (MPECs) but Transiently in Short-Lived Effector Cells (SLECs).

Early T_eff cells can be divided into SLECs (KLRG1^highCD127^low) with terminal differentiation destiny, and MPECs (KLRG1^lowCD127^high) for memory formation. Intriguingly, AhR was highly expressed in MPECs but barely expressed in SLECs (Fig. 3 A and B). Moreover, nuclear AhR (active form) was increased in MPECs (Fig. 3C). Upon entering the nucleus, AhR transcriptionally activates target gene expression. In line with this, the expression of Cyp1a1 and Cyp1b1 were up-regulated in MPECs (Fig. 3D). Consistently, analysis of published RNA-sequencing data (GSE150441) (32) also showed higher AhR expression and activation in MPECs (SI Appendix, Fig. S3A). Previous reports have shown that the differential expression of KLRG1 and CD127 takes around 10 d after stimulation in vivo (33, 34). In our in vivo experiments, the MPEC phenotype was rarely observed in the CD8⁺ OT-I T cell on day 3 or 5 after Lm-OVA infection (SI Appendix, Fig. S3B). However, we found that AhR expression was down-regulated and lost the activity in bulk OT-I cells of mice 3 d after Lm-OVA infection (Figs. 2 B and C and 3E), suggesting that the AhR expression is an earlier event than MPEC appearance. Consistently, mean fluorescence intensity (MFI) of AhR expression exhibited a transient increase 24 h after Lm-OVA infection and then decreased rapidly (Fig. 3F and SI Appendix, Fig. S3C). Notably, we found that AhR expression and activity were necessary for MPEC development, because either AhR knockdown or the use of CH223191 markedly reduced the number of Lm-OVA-induced MPECs in mice (Fig. 3G and SI Appendix, Fig. S3D). In addition, shAhR MPECs also down-regulated the expression of CD62L (Fig. 3H) and expressed lower levels of memory-related transcription factors Tcf7, Lef1, and Bcl6 (Fig. 3I). Similar results were observed in mice treated with CH223191 (SI Appendix, Fig. S3 E and F). In contrast, the activation of AhR in SLECs led to the appearance of the KLRG1^lowCD127^high phenotype (SI Appendix, Fig. S3 G–I). Together, these results suggest that AhR positively regulates MPECs, but negatively regulates SLECs.

Fig. 3.

Activated AhR constitutively exists in MPECs but transiently in SLECs. (A and B) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA. AhR expression in CD8⁺ MPEC (CD127^highKLRG-1^low) and SLEC (CD127^lowKLRG-1^high) from the spleen at day 10 after infection (n = 3). (C) Immunofluorescence microscopy images show the expression of AhR in CD8⁺ MPEC and SLEC (Scale bar: 10 μm.) (n = 3). (D) mRNA levels of Cyp1a1 and Cyp1b1 in splenic MPEC and SLEC were analyzed (n = 3). (E and F) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA. Western blot and flow cytometry analysis of AhR expression in CD8⁺CD45.1⁺ OT-I T cells (n = 3). (G–I) C57BL/6 mice were transferred with 1 × 10⁵ OT-I CD8⁺ T cells transduced with shNC or shAhR retroviruses, followed by infection of Lm-OVA (n = 5 mice per group). (G) Proportions of MPEC and SLEC were analyzed at day 10 post Lm-OVA infection. (H) CD62L expression on OVA-specific MPEC was measured. (I) mRNA levels of Bcl6, Tcf7, and Lef1 in shNC- and shAhR transduced MPEC. Data are representative of ≥3 independent experiments (mean ± SD). P values were calculated using unpaired two-tailed Student’s t test (A, D, G, H, and I) or one-way ANOVA (F). **P < 0.01, ***P < 0.001, ****P < 0.0001.

SLECs Use HIF-1α to Inactivate AhR during the Early Effector Stage.

Next, we investigated the molecular basis by which AhR was down-regulated in SLECs. AhR is known to be complexed with Hsp90, XAP2, p23, and c-Src in the cytoplasm, but translocated into the nucleus upon ligand binding, where it dimerizes with the AhR nuclear translocator (ARNT), also known as hypoxia-inducible factor 1β (HIF-1β), to transactivate gene expression (35). Notably, HIF-1β is also a cellular partner of HIF-1α. Moreover, HIF-1α can be rapidly induced in T_eff cells (36, 37), and the AKT-mTOR pathway activated by TCR signaling up-regulates HIF-1α expression (38). Thus, we assumed that T_eff cells used HIF-1α to inactivate AhR via the occupancy of HIF-1β. Both HIF-1α and HIF-1β were poorly expressed in naive T cells but were gradually induced and reached a peak in T_eff cells after 24 h stimulation (Fig. 4A and SI Appendix, Fig. S4A). By adding 2-NBDG (a fluorescent glucose analog) to activated CD8⁺ T cells for 1 h, we found that 2-NBDG^high T cells (high glycolysis) highly expressed HIF-1α but had low AhR expression and activity, whereas 2-NBDG^low T cells exhibited an inverse expression pattern (Fig. 4 B and C), suggesting that HIF-1α might have an inhibitory effect on AhR. As expected, hypoxia-induced HIF-1α expression inhibited AhR activity in T_eff cells (Fig. 4D); however, AhR activity was enhanced by the addition of the HIF-1α inhibitor KC7F2 (Fig. 4E). In line with this, hypoxia inhibited the formation of IL-15-induced CD8⁺ T_m cells (Fig. 4F). Moreover, in OT-I cell-adoptively transferred mice, we found that AhR expression and activity were enhanced by KC7F2 but inhibited by CoCl₂, a chemical HIF-1α inducer, in the cells upon Lm-OVA infection (Fig. 4 G and H). A similar result was obtained from HIF-1α knockdown (SI Appendix, Fig. S4 B and C). In addition, OT-I-transferred mice were challenged with Lm-OVA, followed by KC7F2 or CoCl₂ treatment once per 2 d. Ten days later, CD8⁺ OT-I T_eff cells were analyzed by co-immunoprecipitation (Co-IP), showing that more AhR bound to HIF-1β by KC7F2 and reduced AhR bound to HIF-1β by CoCl₂ (Fig. 4I). In another setting, we started to treat the mice with KC7F2 or CoCl₂ 10 d later, and on day 30, we analyzed the memory formation, which showed increase and decrease of Tm cells, respectively (SI Appendix, Fig. S4D). These results suggest that following the early induction of AhR, T_eff cells rapidly up-regulate HIF-1α to generate HIF-1α^activeAhR^inactive and HIF-1α^inactiveAhR^active subpopulations, thus giving rise to SLECs and MPECs, respectively. Indeed, we found that high and low levels of HIF-1α were expressed in SLECs and MPECs, respectively, isolated from mice following OT-I cell adoptive transfer and Lm-OVA infection (Fig. 4J). In addition, we stimulated CD8⁺ T cells with anti-CD3/CD28, and found that Ki-67^high T cells highly expressed HIF-1α but exhibited weak AhR activity, whereas Ki-67^low T cells weakly expressed HIF-1α but exhibited strong AhR activity (SI Appendix, Fig. S4 E and F). HIF-1α has been reported to promote AhR proteasomal degradation in type 1 regulatory cells (12). By using cycloheximide (CHX), a protein-synthesis inhibitor, to treat the above 2-NBDG^high CD8⁺ T cells, we found that the half-life of AhR was prolonged by the HIF-1α inhibitor KC7F2 (SI Appendix, Fig. S4G). HIF-1α stability can be enhanced by the oncogenic protein Myc (39). Notably, the activity of Myc in early Teff cells has been reported to be a decision marker of CD8+ T cell memory differentiation (6, 7). In line with this, we found that Myc was highly expressed in Ki-67^high and 2-NBDG^high T cells, but lowly expressed in Ki-67^low and 2-NBDG^low effector T cells (SI Appendix, Fig. S4H). Thus, asymmetric division-generated Myc^high daughter CD8⁺ T cells are likely to contribute to high levels of HIF-1α to inhibit AhR. Together, these results suggest that during naive CD8⁺ T cell priming, differentiated early effector T cells ubiquitously display AhR activity, which can be inactivated by accumulating HIF-1α.

Fig. 4.

SLECs use HIF-1α to inactivate AhR during the early effector stage. (A) Naive CD8⁺ T cells were stimulated with anti-CD3, anti-CD28, and IL-2. HIF-1α expressions were analyzed by western blot (n = 3). (B and C) Naive CD8⁺ T cells were stimulated with anti-CD3, anti-CD28, and IL-2 for 24 h. Flow cytometric analysis of 2-NBDG staining separated cell populations (n = 3). (B) Western blot analysis of AhR and HIF-1α in 2-NBDG^high and 2-NBDG^low CD8⁺ T_eff cells (n = 3). (C) mRNA levels of Cyp1a1 and Cyp1b1 in 2-NBDG^high and 2-NBDG^low CD8⁺ T_eff cells (n = 3). (D) mRNA levels of Cyp1a1 and Cyp1b1 in CD8⁺ T_eff cells cultured in normoxic and hypoxic conditions (n = 3). (E) mRNA levels of Cyp1a1 and Cyp1b1 in CD8⁺ T_eff cells treated with KC7F2 (n = 3). (F) Number of IL-15-induced T_m cells cultured in normoxic and hypoxic conditions (n = 3). (G and H) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA and treatment of KC7F2 and CoCl2. The AhR expression (G) and activation (H) of CD45.1⁺CD8⁺ OT-I T_eff cells were analyzed 10 d post infection. (I) The cell lysate of CD45.1⁺ CD8⁺ OT-I T_eff cells was immunoprecipitated with anti-HIF-1β and blotted with AhR and HIF-1α. (J) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA. HIF-1α expression in splenic MPEC and SLEC was analyzed 10 d after infection. Data are representative of ≥3 independent experiments (mean ± SD). P values were calculated using unpaired two-tailed Student’s t test (C–F) or one-way ANOVA (H). **P < 0.01, ***P < 0.001, ****P < 0.0001.

MPECs Maintain AhR Activity via IL-2.

Next, we investigated how AhR was maintained in its active form in CD8⁺ MPECs. IL-2, a key T cell growth factor mainly produced by activated T cells, is essential for CD8⁺ T cell memory formation (40–43). Moreover, IL-2 signaling has been linked to AhR activity, which activates AhR in CD8⁺ T cells through a STAT5/tryptophan hydroxylase 1 (Tph1)/5-hydoxy tryptophan (5-HTP) dependent pathway (19). We found that IL-2^high CD8⁺ T cells highly expressed AhR, while IL-2^low CD8⁺ T cells poorly expressed AhR (Fig. 5A). Consistently, AhR activity was much higher in the IL-2^high cells, as evidenced by the increased expression of Cyp1a1 and Cyp1b1 (Fig. 5B). Analysis of existing RNA sequencing (RNA-seq) data from CD8⁺ T cells infected with lymphocytic choriomeningitis virus (41) also showed that AhR expression was down-regulated in IL-2⁻ CD8⁺ T_eff and T_m cells (SI Appendix, Fig. S5A). Gene set enrichment analysis (GSEA) based on the previously identified AhR activation gene signature (44) showed that genes transactivated by AhR were enriched in IL-2⁺ T cells (SI Appendix, Fig. S5B). Intriguingly, by determining IL-2 in MPECs and SLECs, we found that MPECs highly expressed IL-2, whereas SLECs poorly expressed IL-2 (Fig. 5C and SI Appendix, Fig. S5C), consistent with a recent study (18). Based on these analyses, we hypothesized that MPECs use IL-2 signaling to maintain an active form of AhR through a STAT5-Thp1-5-HTP pathway. Using high-performance liquid chromatography with mass spectrometry (HPLC–MS), we found that high levels of 5-HTP, the product of Tph1-catalyzed tryptophan, were present in MPECs (Fig. 5D), as well as increased Tph1 expression (SI Appendix, Fig. S5 D and E), compared to SLECs. Moreover, inhibition of STAT5 or Tph1 using STAT5-IN-1 or CP-10188 markedly reduced the number of MPECs, as well as AhR activation (Fig. 5 E and F). In addition, by neutralizing IL-2 with the antibody, we found that 5-HTP production was decreased and AhR activation was abrogated in MPECs (Fig. 5 G and H). Together, these results suggest that IL-2 signaling leads to sustained AhR activation in CD8⁺ MPECs.

Fig. 5.

MPECs maintain AhR activity via IL-2. (A and B) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺ CD8⁺ OT-I T cells, followed by infection of Lm-OVA. (A) AhR expression in IL-2^high and IL-2^low CD8⁺ OT-I T_eff cells was analyzed 7 d after infection (n = 3). (B) mRNA levels of Cyp1a1 and Cyp1b1 in IL-2^high and IL-2^low CD8⁺ T cells (n = 3). (C) Percentage of IL-2-producing OT-I T cells in MPEC and SLEC (n = 6 mice per group). (D) 5HTP abundance in MPEC and SLEC was analyzed by LC/MS (n = 3). (E and F) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺CD45.2⁺CD8⁺ OT-I T cells, followed by infection of Lm-OVA 24 h later. Mice were then treated with STAT5-IN-1 (STAT5 inhibitor) or CP-10188 (Tph1 inhibitor). (E) Proportions of MPEC and SLEC were measured at day 10 post Lm-OVA infection (n = 5 mice per group). (F) mRNA levels of Cyp1a1 and Cyp1b1 in MPEC were analyzed by qPCR. (G and H) MPEC were purified from 10-d infected mice, followed by IL-2 neutralizing antibody treatment for 24 h in vitro (n = 3). (G) 5HTP abundance was measured by LC/MS. (H) mRNA levels of Cyp1a1 and Cyp1b1 were analyzed. Data are representative of ≥3 independent experiments (mean ± SD). P values were calculated using unpaired two-tailed Student’s t test (A–D, G, and H) and or one-way ANOVA (E and F). *P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

AhR Licenses CD8⁺ MPECs in a Quiescent State for Memory Formation.

Next, we investigated the mechanism by which AhR regulates memory development in CD8⁺ T cells. The pivotal features of T cell memory are quiescence and longevity. Given that AhR activation not only mediates cell dormancy (45, 46) but also up-regulates inhibitory receptors such as PD-1 and LAG3 in CD8⁺ T cells (15, 19), which are required for memory homeostasis and their deficiency impairs the T cell memory pool, we speculated that AhR regulates the quiescent state of CD8⁺ T cells in order to favor memory formation and survival. Quiescent CD8⁺ T_m cells use oxidative phosphorylation rather than glycolysis as their metabolic mode (47). Using Seahorse analysis, we found that AhR inhibition markedly elevated glycolysis and reduced oxidative phosphorylation (OXPHOS) in CD8⁺ MPECs, as evidenced by the increased extracellular acidification rate (ECAR) and decreased oxygen consumption rate (OCR) with impaired SRC (Fig. 6 A and B). Similar results were observed in IL-15 differentiated CD8⁺ T_m cells treated with CH223191 (SI Appendix, Fig. S6 A–C). In addition, ¹³C-glucose tracing indicated that AhR inhibition elevated glycolysis flux, as evidenced by the increase in pyruvate and lactate in MPECs (Fig. 6C), suggesting that AhR promotes the quiescent state of CD8⁺ MPECs. In addition, we found that AhR-deficient CD8⁺ T_m cells consumed more glucose than WT T_m cells (SI Appendix, Fig. S6D), and the key glycolytic enzyme, hexokinase 2 (HK2) (48), was up-regulated in the MPECs of mice treated with CH223191 (Fig. 6D), as well as in AhR-deficient CD8⁺ T_m cells (SI Appendix, Fig. S6E). Given that HIF-1α up-regulates glycolytic enzymes, including HK2 (49) and AhR deficiency up-regulates HIF-1α expression (12), we speculated that CD8⁺ MPECs switched the glycolysis to OXPHOS by using AhR to antagonize HIF-1α. Coincidently, two xenobiotic responsive elements (XREs) were observed in the promoter region of HIF-1α. ChIP-qPCR analysis showed that increased AhR bound to the HIF-1α promoter in CD8⁺ T_m cells compared to that in CD8⁺ T_eff cells (Fig. 6E). However, both luciferase assay and qPCR showed that AhR promoted mRNA transcription of HIF-1α (Fig. 6 F and G). Despite this, we found that AhR activation shortened the half-life of HIF-1α in CD8⁺ T_eff cells (Fig. 6H). Consistently, HIF-1α ubiquitination was enhanced by AhR activation (Fig. 6I). In addition, in the AhR-deficient CD8⁺ T_m cells, HK2 up-regulation could be reversed by HIF-1α knockdown, suggesting CD8⁺ MPECs use AhR to antagonize HIF-1α and promote the switch from glycolysis to OXPHOS (Fig. 6J). Inhibitory receptors have been reported to play important roles in CD8⁺ T cell memory formation (17, 18). We found that AhR inhibition also resulted in down-regulation of PD-1, LAG-3, and Tim-3 in CD8⁺ MPECs (SI Appendix, Fig. S6F). IL-15-derived AhR^−/− CD8⁺ T_m cells exhibited lower levels of IRs than WT CD8⁺ T_m cells (SI Appendix, Fig. S6G). In addition, we analyzed the cell cycle, and as expected, 63.8% of MPECs were in the G0/G1 phase, which was abrogated by AhR inhibition (Fig. 6K). AhR inhibition also resulted in a global reduction of quiescence regulators such as Klf2, Klf6, Gimap5, Btg1, Btg2, and VISTA in MPECs (SI Appendix, Fig. S6H). Together, these results suggest that AhR licenses CD8⁺ T_m cells in a quiescent state.

Fig. 6.

AhR licenses CD8⁺ MPECs in a quiescent state for memory formation. (A and B) C57BL/6 mice (CD45.2⁺) were transferred with 1 × 10⁵ CD45.1⁺CD8⁺ OT-I T cells, followed by infection of Lm-OVA 24 h later. Mice were then treated with CH223191 every day. MPECs were purified from 10-d infected mice, and ECAR (A) and OCR (B) of MPEC were analyzed (n = 3). (C) MPECs were purified from 10-d infected mice, and then cultured in ¹³C-glucose medium for 24 h. ¹³C-labeled lactate and pyruvate were measured by LC/MS (n = 3). (D) Mice were treated as (A), the expressions of HK2 in MPECs were analyzed by western blot (n = 3). (E) ChIP-qPCR analysis was performed with an antibody to AhR and HIF-1α-promotor-specific primer (n = 3). (F) NIH3T3 cells were cotransfected with a HIF-1α promoter-luciferase reporter PGL4.10 and Flag-AhR plasmid for 24 h, followed by analysis of luciferase activity (n = 3). (G) mRNA levels of HIF-1α in CD8⁺ T_eff cells in the presence or absence of FICZ were analyzed (n = 3). (H) Western blot analysis of AhR expression in CD8⁺ T_eff cells treated with CHX (10 μg/mL) in the presence or absence of FICZ (10 μM) (n = 3). (I) Activated CD8⁺ T_eff cells were treated with FICZ or Kyn for 48 h. The cell lysates were immunoprecipitated with an anti-HIF-1α antibody and immunoblotted with anti-ubiquitination antibody (n = 3). (J) HIF-1α in AhR^−/− Tm cells was knocked out through the electroporation of Cas9-sgRNA ribonucleoproteins (RNPs). The expressions of AhR and HK2 in WT, AhR^−/− or AhR^−/− HIF-1α-KO T_m cells were analyzed by western blot (n = 3). (K) MPECs purified from 10-d infected mice were assayed for cell cycle analysis. Data are representative of ≥3 independent experiments (mean ± SD). P values were calculated using unpaired two-tailed Student’s t test (A–E, G, and K) and or one-way ANOVA (F). *P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

Discussion

CD8⁺ memory T (T_m) cells are essential components of long-lived immunity against infections and cancers. Despite the recently addressed long-standing debate on the origin of T_m cell differentiation from effector or naive cells, a fundamental question has been raised regarding how a minority of effector cells are selected from the effector pool and give rise to T_m cells. Recent studies have revealed the involvement of AhR in the differentiation of intestinal and skin CD8⁺ tissue-resident memory T cells (50–52). In this study, we found that AhR acts as a crucial transcriptional factor for CD8⁺ T_m cells in various locations, including the spleen, lymph nodes, blood, and peripheral tissues such as the lungs and liver. We provided evidence that 1) TCR signaling triggers ROS production to rapidly induce AhR in early CD8⁺ T_eff cells via an Nrf2-dependent pathway; 2) however, upon T cell priming and proliferation, HIF-1α can accumulate in short-lived effector cells, which inactivates AhR via competitive inhibition by binding HIF-1β; and 3) unlike SLECs, MPECs use IL-2 to maintain AhR activity, and in return, AhR confers CD8⁺ T_eff cells in a quiescent state for memory formation. This molecular mechanistic elucidation provides insight into how early CD8⁺ effector T cells are selected from the effector pool and differentiate into T_m cells (SI Appendix, Fig. S6I).

Both HIF-1α and AhR systems have evolved from ancient multicellular organisms and respond to alterations in O₂ and ROS (53, 54). Under hypoxia, HIF-1α promotes glycolysis. Under oxidative stress, AhR is up-regulated and augments the indoleamine 2,3-dioxygenase (IDO) signaling pathway. Thus, AhR and HIF-1α may interact physiologically to regulate different cell behaviors and states. On the one hand, they mutually antagonize each other (55); however, they also cooperate to regulate cell metabolic fitness, including glycolysis (56).

Although we found that early T_eff cells select AhR as a turning point to differentiate toward either terminal T_eff or T_m cells, based on the dominance between AhR and HIF-1α, how one of the two daughter cells generated from the same parental T_eff cell is destined to activate HIF-1α is not well addressed in this study. While extrinsic factors, such as partial pressure of oxygen, may affect HIF-1α expression, intrinsic factors are undoubtedly pivotal. Recent studies have highlighted that asymmetric cell division plays an important role in T cell differentiation and development (6, 7). It has been reported that mTORC1 activity is asymmetrically inherited after the first cell division of naive CD8⁺ T cells (57). Notably, mTORC1 may facilitate asymmetric c-Myc synthesis in CD8⁺ T cells (58), and high c-Myc combined with high cBAF can bias differentiation toward an effector fate (6). These findings may be consistent with the induction of HIF-1α in SLECs reported in this study, considering that c-Myc can stabilize HIF-1α and enhance HIF-1α activity (39, 59, 60). Based on our findings and others, we propose the following molecular events for early T cell memory fate: 1) upon TCR engagement with MHC-I-antigenic peptide, CD8⁺ T cells initiate asymmetric division with asymmetric distribution of c-Myc in the divided sister cells; 2) TCR signaling drives the increase in ROS (30), which triggers the activation of AhR in the cells; 3) early T_eff cells with high c-Myc activity up-regulate HIF-1α expression to inhibit AhR for effector cell fate; and 4) early divided T_eff cells with low c-Myc activity are licensed to enter a quiescent state driven by the activated AhR for memory fate. Overall, the elucidation of molecular basis for memory formation from early CD8⁺ Teff cells has therapeutic implications by translating into more efficient vaccines and T cell–based immunotherapy against cancer and infections.

Materials and Methods

Animals.

Female WT C57BL/6 mice that were 6 to 8 wk old were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. OT-I TCR-transgenic mice [C57BL/6-Tg(TcraTcrb)1100Mjb/J] were a gift from H. Zhang (Sun Yat-sen University, China). AhR^−/− mice were presented by Jun Yan (Third Military Medical University), and OT-I mice and CD45.1 mice were crossed to obtain CD45.1⁺OT-I mice. These animals were maintained in the Animal Facilities of Tongji Medical College, Huazhong University of Science & Technology under pathogen-free conditions.

Animal Experiments and Treatment Protocol.

For the infection model, 1 × 10⁴ CFUs of Lm-OVA were administrated into 6- to 8-wk-old female AhR^−/− and WT C57BL/6 mice intravenously, bacterial loads in the spleen and liver were analyzed 48 h after infection. OVA-specific CD8⁺ T cells in the spleen, lymph nodes, blood, and lungs were analyzed by flow cytometry on day 7. On day 30, the mice were re-challenged with 1 × 10⁶ CFU Lm-OVA, and the bacterial burdens in the spleen and liver were analyzed. For the OT-I transfer experiment, 1 × 10⁵ shAhR-CD8⁺ T cells or shNC-CD8⁺ OT-I T cells were transferred into 6-8-week-old female CD45.2 recipients. One day after OT-I cell transfer, 5 × 10⁵ CFUs of Lm-OVA were administrated intravenously. In some studies, 1 × 10⁵ CD8⁺ Tn cells isolated from CD45.1⁺ OT-I mice were transferred into 6- to 8-wk-old female CD45.2 recipients. Ten mg/kg CH223191(Selleck, S7711), 5 mg/kg STAT5 inhibitor (STAT5-IN-1; Selleck), or 5 mg/kg Tph1 inhibitor (CP-10188, Selleck) were administrated intraperitoneally everyday according to the experimental needs.

Analysis or sorting of MPECs (CD127^high KLRG1^low) and SLECs (CD127^low KLRG1^high) was conducted from the spleen of mice infected with Lm-OVA on day 10, which had been previously transferred with CD8⁺ OT-I T cells.

CD8⁺ Memory T Cell Generation.

OT-I splenocytes were activated with OVA_257-264 peptides (200 ng/mL) and IL-2 (10 ng/mL) for 3 d. The cells were subsequently collected, washed three times with RPMI-1640 medium, and further cultured in the presence of IL-15 (10 ng/mL) for an additional 4 d to program CD8⁺ T_m cells.

Retroviral Transduction.

For retroviral transduction experiments, OT-I splenocytes were activated for 24 h and then transduced with concentrated retrovirus carrying pROV-U6-shAhR-EF1A(S)- EGFP, or scramble shRNA. EGFP was used as a marker of retroviral expression. Retroviruses were spin-inoculated at 450g for 2 h at 32 °C in media containing hexadimethrine bromide (5 μg/mL) and IL-2 (10 ng/mL). For adoptive-transfer experiments, EGFP⁺ cells were sorted using a flow cytometer 24 h after retroviral transduction.

Flow Cytometry.

For analysis of surface markers, single-cell suspensions were stained with following fluorochrome-conjugated antibodies following established guidelines (61): FITC-conjugated anti-CD45.1 (Biolegend, 110705), PerCP-Cyanine5.5-conjugated anti-CD8 (Biolegend, 110733), Pacific Blue-conjugated anti-CD8 (Biolegend, 100725), PE-conjugated anti-CD69 (Biolegend, 104508), FITC-conjugated TIM3 (Genetex, GTX54055), APC-conjugated anti-LAG3 (Biolegend, 125210), APC-conjugated anti-PD-1 (Biolegend, 135210), APC-conjugated anti-IL-2 (Biolegend, 503809), APC-conjugated anti-CD25 (Biolegend, 101909), FITC-conjugated anti-IFN-γ (Biolegend, 505805), APC-conjugated anti-TNF-α (Biolegend, 506307), PE-conjugated anti-H-2 K(b)/SIINFEKL Tetramer (Helixgen, HG08T14028), BV421-conjugated KLRG1 (Biolegend, 138413), BV605-CD127 (Biolegend, 135025), and APC-Cy7-conjugated anti-CD62L (Biolegend, 104427), For intracellular staining, cells were first fixed with IC Fixation Buffer (eBioscience 00-8222-49) and permeabilized with Permeabilization Buffer (Invitrogen, 00-8333-56). Flow cytometry was performed using Accuri C6 or Sony ID7000 system and analyzed with FlowJo software.

Western Blot Analysis.

Equal amounts of protein of each sample were run on an SDS-PAGE gel and transferred to nitrocellulose. Nitrocellulose membranes were blocked in 5% bovine serum albumin (BSA) and probed with antibodies overnight, antibodies and dilutions were as follows: anti-AhR, 1:1,000 (Biolegend, 694504); anti-Nrf2, 1:1,000 (CST, 12721); anti-HIF-1α 1:2,000 (Proteintech, 20960-1-AP); anti-HIF-1β 1:1,000 (Proteintech, 14105-1-AP); anti-HK2, 1:1,000 (CST, 2867S); anti-Tph1, 1:1,000 (CST; 12339S); anti-LDHA, 1:1,000 (CST; 3582S); anti-c-Myc 1:1,000 (Proteintech, 67447-1-Ig); β-actin, 1:10,000 (Proteintech, 81115-1-RR). Secondary antibodies conjugated to horseradish peroxidase were followed by enhanced chemiluminescence (Thermo Fisher).

Statistical Analysis.

Statistical analysis was performed with Prism 8.0 (GraphPad) by two-tailed paired Student’s t test, or one-way ANOVA with Dunnett’s multiple comparisons test. Graphs show individual samples, and center values indicate mean. P values of less than 0.05 are considered statistically significant (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns: not significant, P > 0.05).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.