New reports provide further insight into the role of the transcription factor BATF in pivoting the differentiation of CD8+ T cells away from T cell exhaustion and facilitating the transition of these cells into potent effectors.
T cell exhaustion is a complex developmental program that results in the progressive loss of T cell effector function in response to environments of chronic antigenic stimulation. While this “dysfunctional” state probably occurs to protect the host against the pathophysiology of a sustained T cell response1, this programming remains a hurdle to effective cellular therapies, particularly in the context of cancer. With the exciting discovery of a “stem-like” progenitor population within the exhausted T cell pool, insight into which regulators can facilitate the transition of progenitors to more terminally differentiated populations remains an open area of exploration. In this issue of Nature Immunology, armed with high-dimensional analyses examining the gamut of transcriptional and epigenetic profiles among the cells, studies by Chen et al.2 and Seo et al.3 demonstrate that early manipulation of BATF mediates the transition of CD8+ T cells away from exhaustion and skews these cells toward an effector phenotype endowed with cytolytic function in chronic viral infections and tumors (Fig. 1).
Recent work has demonstrated that the pool of exhausted T cells is heterogeneous, with several distinct subtypes of T cells represented in various developmental states. Specifically, the pool was previously thought to comprise terminally exhausted CD8+ T cells, which predominantly express high amounts of inhibitory receptors and display poor proliferative potential in conjunction with impaired cytotoxic capacity. Nuanced work has now revealed that the pool also contains a relatively small population of exhausted progenitor T cells, marked by an enriched “stem-like” signature. More recently, a population of CX3CR1+ effector T cells4,5 has been described, which may represent a transitory population (Fig. 1). However, whether this heterogeneity represents a developmental spectrum and what the specific determinants regulating these processes are remain active areas of investigation. Moreover, discovering regulatory processes that control or mediate this transition may provide new avenues for improving cell-based therapeutics.
As the Cui lab had previously demonstrated5 the presence of CX3CR1+ effector cells within the population of CD8+ T cells responding to chronic lymphocytic choriomeningitis virus (LCMV), this group hypothesized that these three subsets within the T cell pool (terminally exhausted (CX3CR1–Ly108–), progenitor (Ly108+) and CX3CR1+ effector) collectively contribute to antigen control. Formation of the CX3CR1+ effector population was suggested to be in part due to CD4+ T cell help, which would enable the bifurcative transition of T cells from the progenitor population into either CX3CR1+ effector or terminally exhausted T cells (Fig. 1). Thus, Chen et al.2 sought to define the transcriptional states, chromatin landscapes and histone patterns of each subpopulation and, ultimately, to identify transcription factors that may regulate this differentiation process.
Utilizing single-cell gene expression data from antigen-specific CD8+ T cells isolated from acute or chronic LCMV infection, Chen et al.2 determined by SCENIC and UMAP analyses that the three distinct CD8+ T cell populations were present. These populations, found among LCMV-specific T cells during chronic infection, had distinct regulon and gene expression signatures. While many of the genes identified within each cluster (such as Tcf7 in the progenitor population and Eomes and Nr4a in exhausted T cells) confirmed prior studies, Chen et al.2 did note an unexpected enrichment both in NF-κB-family regulons in Ly108+ cells and in several factors expressed in CX3CR1+ cells, such as Tbx21, Zeb2, Bhlhe40, Klf2 and Klf3. Moreover, Chen et al.2 noted that, while there were some differences identified only in cells isolated from mice with chronic LCMV infection, in the expression of genes such as Tox (a key regulator of T cell exhaustion), there was gene expression overlap between short-lived effector cells (acute) and CX3CR1+ (chronic) populations and between memory-precursor effector cells (acute) and Ly108+ (chronic) progenitor cells. Furthermore, short-lived effector cells and CX3CR1+ cells were also found to have similar enhancer profiles by ATAC–seq (assay for transposase-accessible chromatin using sequencing), suggesting that these cells are not only transcriptionally but also epigenetically analogous. Interestingly, these collective data therefore support a model in which CD8+ T cell populations initially undergo similar developmental processes in both acute and chronic infection but then specific adaptations are subsequently acquired in the face of chronic antigen.
Intriguingly, when the three subsets were analyzed by ATAC–seq, Chen et al.2 noted that the top two transcription factors enriched within the CX3CR1+ effector subset as compared to the exhausted population were T-bet and BATF. When the authors deleted Tbx21, which encodes T-bet, in the context of a bone-marrow-chimera experiment, deficiency in T-bet led to a diminished presence of CX3CR1+ effector cells, increased expression of inhibitory receptors and EOMES on these cells and resulted in weakened viral titer control. Haploinsufficiency of BATF post-activation led to an overall impaired generation of CX3CR1+ effector cells, with subsequent increases in the exhausted population. Furthermore, this deficiency was associated with compromised cytolytic functions and viral control. However, this pronounced decrease in CX3CR1+ effector T cells was only noticeable at later time points during chronic infection, as examination of early time points did not reveal differences in the frequency nor viral titer. Chromatin accessibility analysis of cells that had wild-type, null or haploinsufficient BATF expression demonstrated that ~4,000 enhancer regions were lost in response to BATF deficiency—a loss mainly found in regions of effector function–related genes. Further analysis demonstrated that BATF was responsible for a significant percentage of active enhancer regions in genes such as Ifng, Gzmb, Klf2, Txb21, Cx3cr1 and Prf1 (Fig. 1).
Along similar investigative lines, new work coming out of the Hogan lab also supports the involvement of BATF in curbing the differentiation state of chimeric antigen receptor (CAR) T cells toward a more effector-like state. Along with several other labs, Seo et al.3 demonstrated that TOX, in conjunction with NR4A transcription factors, regulates T cell exhaustion6,7,8. As exhaustion programming results in decreases in both the expression and accessibility of AP-1/bZIP motifs, Seo et al.3 hypothesized that restoring AP-1 complexes would counteract exhaustion. Accordingly, BATF was one hit identified during their screen for factors that enhanced NFAT–AP-1 activity. Seo et al.3 subsequently serially tested whether these hits enhanced the antitumor activity of CAR T cells in a murine model of melanoma expressing human CD19. Surprisingly, it was the BATF-overexpressing (BATF-OE) CAR T cells that resulted in the most significant reduction in tumor size and extended survival as compared to a control vector. Strikingly, transfer of BATF-OE CAR T cells after tumor establishment still resulted in decelerated tumor outgrowth (Fig. 1). Characterization of tumor-infiltrated BATF-OE CAR T cells demonstrated a global reduction in the expression of several exhaustion-associated markers, such as TOX, NR4A members and EOMES. Furthermore, there was a notable increase in the frequency and number of CAR T cells with BATF-OE, which correlated with increased Ki67 expression, primarily within cells lacking TCF-1 expression—findings that were subsequently confirmed using CyTOF (cytometry by time of flight). Likewise, CyTOF revealed that BATF-OE in CAR T cells increased the expression of effector-related markers, such as ICOS, granzyme B and CD25, which suggested a skewing of differentiation toward an effector-like T cell that was also observed following the loss of a PD-1hiTOXhi population. Moreover, this increase in effector-like tumor-infiltrating leukocytes resulted in tumoral rejection and promoted a population that was capable of eliciting protection against tumor recurrence.
Prior work demonstrated that BATF cooperates with IRF4, therefore Seo et al.3 next sought to determine whether interaction with IRF4 was necessary for the observed effects in BATF-OE CAR T cells. Using a mutated form of BATF that prevents BATF–IRF4 interaction, it was found that CAR T cells transduced with this construct were unable to control tumoral outgrowth, resulting in reduced survival. Similarly, transfer of mutated-BATF or BATF-deficient CAR T cells into established tumors also resulted in a blunted capacity to control the tumor, which was associated with decreased frequency and number of infiltrated CAR T cells. Transcriptional and epigenetic analyses of the T cells demonstrated that expression of Tbx21 and Eomes is altered early after activation in BATF-OE CAR T cells, which may account for the propensity of such cells to enhance an effector-like state at the expense of an exhaustion-associated signature.
Interestingly, prior data suggested that BATF was in part responsible for limiting the functionality of exhausted T cells, and prior reports demonstrated that BATF expression was increased in exhausted T cells and that knockdown of BATF in HIV-specific T cells improved effector function9,10. However, recent studies have implicated BATF in the early bifurcative process of progenitor versus effector T cell differentiation, as BATF deficiency resulted in increases in TCF-1+TIM3– (progenitor) cells at the expense of TCF-1–TIM3+ (effector) cells at time points as early as 5 days post-infection with chronic LCMV11. Furthermore, modulation of factors targeting BATF, such as deleting Regnase-1, has also been noted to result in improved antitumor efficacy, in part by promoting increased BATF expression12,13. In line with these data, the exciting studies reported herein suggest that BATF, in conjunction with IRF4, has the capacity to skew T cells away from exhaustion-like programming and favors the formation of a more effector (CX3CR1+)-like T cell under conditions of chronic infection (Fig. 1). Subsequently, BATF falls in line with a growing list of transcription factors that are being targeted in an effort to bid “hasta la vista” to T cell exhaustion.
Blank, C. U. et al. Nat. Rev. Immunol. 19, 665–674 (2019).
Chen, Y. et al. Nat. Immunol. https://doi.org/10.1038/s41590-021-00965-7 (2021).
Seo, H. et al. Nat. Immunol. https://doi.org/10.1038/s41590-021-00964-8 (2021).
Hudson, W. H. et al. Immunity 51, 1043–1058.e4 (2019).
Zander, R. et al. Immunity 51, 1028–1042.e4 (2019).
Seo, H. et al. Proc. Natl Acad. Sci. USA 116, 12410–12415 (2019).
Alfei, F. et al. Nature 571, 265–269 (2019).
Khan, O. et al. Nature 571, 211–218 (2019).
Man, K. et al. Immunity 47, 1129–1141.e5 (2017).
Quigley, M. et al. Nat. Med. 16, 1147–1151 (2010).
Utzschneider, D. T. et al. Nat. Immunol. 21, 1256–1266 (2020).
Wei, J. et al. Nature 576, 471–476 (2019).
Zheng, W. et al. Blood https://doi.org/10.1182/blood.2020009309 (2021).
Chen, Z. et al. Immunity 51, 840–855.e5 (2019).
This work was supported by the National Institutes of Health R01AI114442 and R01CA237311 to B.Y. and F32CA250155-01A1 to S.K.B. and by the American Lebanese Syrian Associated Charities (ALSAC) to B.Y. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
B.Y. has patents related to epigenetic biomarkers and methods for enhancing CAR T cell function. S.K.B. and X.L. declare no competing interests.
About this article
Cite this article
Boi, S.K., Lan, X. & Youngblood, B. BATF targets T cell exhaustion for termination. Nat Immunol 22, 936–938 (2021). https://doi.org/10.1038/s41590-021-00978-2