How transcription factors drive choice of the T cell fate

Abstract

Recent evidence has elucidated how multipotent blood progenitors transform their identities in the thymus and undergo commitment to become T cells. Together with environmental signals, a core group of transcription factors have essential roles in this process by directly activating and repressing specific genes. Many of these transcription factors also function in later T cell development, but control different genes. Here, we review how these transcription factors work to change the activities of specific genomic loci during early intrathymic development to establish T cell lineage identity. We introduce the key regulators and highlight newly emergent insights into the rules that govern their actions. Whole-genome deep sequencing-based analysis has revealed unexpectedly rich relationships between inherited epigenetic states, transcription factor-DNA binding affinity thresholds and influences of given transcription factors on the activities of other factors in the same cells. Together, these mechanisms determine T cell identity and make the lineage choice irreversible.

Figure 1 |

Schematic of early T cell developmental stages in mice. Stages of development within the mouse thymus are shown up to TCRαβ-dependent positive selection, including the developmental checkpoints and major cell surface phenotype markers. Top: phases of responsiveness to indicated growth and survival signals from the environment. Middle: Approximate timing of TCR gene rearrangements is shown below the cells; “TCRβ gene rearrangement” indicates the stage when V-DJβ rearrangement can produce a complete TCRβ chain. Expression of genes encoding RAG1 and RAG2 recombinases and CD3 components is also depicted. Blue shading shows stages after commitment to the T cell lineage, as defined by loss of ability to generate non-T cells when placed in an alternative lineage-promoting environment. This coincides with expression of BCL11B at single-cell level. Bottom: main alternative fates accessible to the indicated pre-commitment cell types if they are withdrawn from Notch signalling. The ability to generate B cells is apparently confined to the most immature ETPs; cells through DN2 stage are also reported to generate mast cells and macrophages under Notch withdrawal conditions (not depicted), and commitment timing is considerably earlier for fetal thymocytes than for postnatal thymocytes (reviewed in ^,).

Figure 2 |

Major changes in epigenetic state and transcription factor expression in mouse pro-T cells. The figure summarizes changes in genomic accessibility patterns and patterns of transcription factor expression based on data in REFS^, and REFS^,, respectively. Indicated levels approximate a logarithmic scale. Labels indicate RNA expression except for PU.1 (p), which designates PU.1 protein. Whereas Spi1 RNA (encoding PU.1) is expressed like HHEX and BCL11A (not shown), the PU.1 protein persists longer due to its high stability. Note the overlap in expression of progenitor transcription factors (such as LYL1 and PU.1 protein) and T cell lineage specification transcription factors (such as TCF1 and BCL11B) in late ETPs, DN2a cells and DN2b cells. This overlap extends through multiple cell cycles and has been validated at the single-cell level.

Figure 3 |

Conditionality of transcription factor binding at genomic sites. Schematic illustrations of how the same transcription factor may differentially occupy genomic sites based on their intrinsic affinities for binding by the factor, their chromatin accessibility status, and their comparative advantage when a second transcription factor with its own binding specificities can interact with the first factor. Schematics in (A-D) are drawn from examples in REFS^,,. (A) Default occupancy patterns for an idealized transcription factor on six sites that it recognizes with different intrinsic affinities, at different expression levels of the transcription factor. (B) Alterations of the default occupancy pattern in a cell type where some sites are occluded by closed chromatin. This part of the figure schematizes results seen for PU.1 in pro-T cells. Sites in closed chromatin may still be bound at high transcription factor concentrations if they have high-affinity motifs. (C) Cooperative recruitment: the ability of a potential interaction partner (light magenta) to enhance occupancy of marginal sites by the main transcription factor by coordinated binding. (D) Cofactor “theft”: loss of binding by the main transcription factor from a subset of occupancy sites (“sensitive sites”), observed when certain partners (blue) recruit it to some alternative site(s). The same transcription factor can have either role in different contexts. (E, F) Impacts of the mechanisms described on the actual patterns of occupancy by RUNX1 before (E) and after (F) T cell lineage commitment. (E) Biased overlap of pre-commitment pattern of RUNX1 with sites occupied by PU.1 (data from REF). (F) Extensive interaction of sites occupied by RUNX1 with BCL11B binding after commitment (data from REF).

Figure 4 |

Transcription factor binding changes at key developmentally regulated loci. Summary schematics are shown for transcription factor occupancies observed by ChIP-seq at indicated loci before T cell lineage commitment (in ETPs (DN1 cells)) and after T cell lineage commitment (in DN2b and DN3 cells), comparing PU.1 in ETP samples, BCL11B in DN2b samples, E2A in DN3 samples (Rag2-knockout thymocytes), and GATA3 and RUNX1 in both ETPs and DN2b–DN3 samples. Original data were from REFS^,,,, aligned after re-mapping to the mm10 build of the mouse genome. Cd3gde cluster genes (panel A), Ets1 (panel C), and Rag1-Rag2 (panel D) are upregulated sharply from DN2a to DN2b stages, whereas the progenitor cell regulatory gene Meis1 (panel B) is downregulated before T cell lineage commitment. Genomic regions depicted are shown at the bottom of each panel in mm10 coordinates, and transcription factor binding positions are shown to scale. Smaller arrows indicate low detected occupancy of the indicated transcription factor. Note the changes in binding patterns of GATA3 and RUNX1 from pre- to post-commitment despite modest changes in expression. Bifunctional transcription factor RUNX1 undergoes single-site changes in occupancy at some loci but multi-site increases in occupancy at others, suggesting that its binding is regulated by broader genomic domain opening.