Asynchronous combinatorial action of four regulatory factors activates Bcl11b for T cell commitment

Abstract

During T cell development, multipotent progenitors relinquish competence for other fates and commit to the T cell lineage by turning on Bcl11b, which encodes a transcription factor. To clarify lineage commitment mechanisms, we followed developing T cells at the single-cell level using Bcl11b knock-in fluorescent reporter mice. Notch signaling and Notch-activated transcription factors collaborate to activate Bcl11b expression irrespectively of Notch-dependent proliferation. These inputs work via three distinct, asynchronous mechanisms: an early locus 'poising' function dependent on TCF-1 and GATA-3, a stochastic-permissivity function dependent on Notch signaling, and a separate amplitude-control function dependent on Runx1, a factor already present in multipotent progenitors. Despite their necessity for Bcl11b expression, these inputs act in a stage-specific manner, providing a multitiered mechanism for developmental gene regulation.

Figure 1. Bcl11b expression turns on in developing T-cell progenitors, and stays on throughout development and maturation

a) Flow cytometry analysis of Bcl11b-YFP levels in CD4, CD8 double negative (DN) thymocytes from single mice containing one, two or zero copies of the Bcl11b-YFP reporter allele, showing gates used for defining DN sub-populations (left), and Bcl11b-YFP level distributions for those populations (right). b) Mean Bcl11b-YFP levels from flow cytometry measurements in various T-cell populations from the thymus (left) or spleen (right) in Bcl11b-YFP reporter mice. Baseline levels in B cells and natural killer (NK) cells, both of which of which do not express Bcl11b, do not exceed 2.6 on this scale, similar to that of T cells from control C57Bl/6 mice. c) Flow cytometry analysis of T-cell developmental progression and Bcl11b-YFP levels in sorted DN progenitors, cultured for three days on OP9-DL1 monolayers with 5 ng/ml IL-7 and Flt3L. Data are representative of two independent experiments (a,b), or three independent experiments (c).

Figure 2. Bcl11b activation coincides with T-cell commitment

a) Analysis of dendritic cell (DC, CD11c⁺) or NK cell (NK1.1⁺ or DX5⁺, mixed antibodies) potential in thymic DN progenitors, assayed after 7–8 days on OP9 monolayers with IL-7 and Flt3L. Wells were seeded with 300 cells (top), or the number indicated (bottom). Bars: fraction of wells with DC or NK progeny for the given number of seeded cells, and inferred lineage potential. Reduced NK and DC potentials between Bcl11b-YFP⁻DN2 and Bcl11b-YFP⁺DN2 cells are significant (* - p = 0.01; ** - p = 0.002; z-test on logistic fit). b)-f) Transcriptomic analysis of bone-marrow DN2 progenitors using RNA-Seq. b) Sort strategy; c) Analysis of developmental potential in bone-marrow DN progenitors by limiting dilution assays, showing less NK potential in Bcl11b-YFP⁺DN2 than in Bcl11b-YFP⁻DN2 (* - p = 0.01; z-test on logistic fit). d) Principal component (PC) analysis (solid circles), showing projections along first two PC axes (top), and loadings of the top ten PC genes (bottom). DN populations from our previous study are shown (open circles). Arrow represents T-cell developmental trajectory. e) Hierarchical clustering analysis of the expression of genes differentially expressed in Bcl11b-YFP^off and Bcl11b-YFP^on DN2 cells (left, q < 0.005), and selected T-cell regulatory genes (right). See Supplementary Tables 1 and 2 for FPKM values. Data in (a) represent 12, 11, and 12 replicates (for 2, 10 and 50 cells respectively) in one experiment, with similar results for three independent experiments. Data in (c) represent 10 replicates in one experiment, with similar results for two independent experiments. Transcriptomic data (d,e) represent biological replicates from two experiments.

Figure 3. Notch and IL-7 play opposing, stage-specific roles during early T-cell development

a) Flow cytometry analysis of Bcl11b-YFP expression for different BM-derived DN progenitor populations, after culture on OP9 or OP9-DL1 for the indicated number of days. Initial Bcl11b-YFP levels are shown (dotted lines). b) Flow cytometry of different BM-derived DN progenitor populations, after culture under the indicated conditions for 4 days. Results are representative of two (IL-7 titration) or five (Notch withdrawal) independent experiments.

Figure 4. Bcl11b activation is not coupled to Notch-dependent cell proliferation

a) Flow cytometry analysis of Bcl11b expression and cell division kinetics for Bcl11b-YFP⁻DN2 progenitors stained with the cell division dye CellTrace Violet and cultured for four days. CellTrace Violet levels in starting cells are shown (dotted lines). The number of cell divisions is given by the log ratio of initial to final CellTrace Violet levels. b–c) Flow cytometry analysis of BM DN2 progenitors transduced with c-Myc, or treated with the cyclin-dependent kinase (CDK) 4/6 inhibitor PD0332991 (2.1 µM), and analyzed after four days. b) Fold increase in cell numbers under the indicated conditions. c) Analysis of Bcl11b-YFP, CD25 and CD44 expression in the indicated cultured populations. EV – empty vector. Data in (a) represent mean and S.D. of three replicates from one experiment, with similar results seen in independent experiments. Bars in (b) represent the mean of two replicates, each shown with circles; results in (b,c) are representative of two independent experiments.

Figure 5. Notch signaling increases the probability of all-or-none Bcl11b activation

a)–d) Long-term timelapse imaging analysis of Bcl11b activation dynamics in single Bcl11b-YFP⁻ DN2 progenitors on OP9 or OP9-DL1 monolayers. a) Time evolution of Bcl11b-YFP levels for a cohort of cells on OP9-DL1 (left, top) or OP9 (left, bottom) monolayers. Histograms (right) give binned cell data obtained at the indicated times (top). Data were shown starting at t~25 hrs onwards, reflecting the onset time for mCherry expression. Smooth lines indicate best fits to mixed Gaussian distributions. b) Single-cell Bcl11b-YFP time traces on OP9-DL1 (left) or OP9 (center) monolayers. Traces where Bcl11b levels rise above background (gray shading) are indicated in red. Bars give fraction of lineages turning on Bcl11b, showing significant increase in Bcl11b-YFP⁺ lineages on OP9-DL1 (χ²=3.25, d.f. = 1; p=0.072). c) Bcl11b-YFP⁺ cells at 40 and 80 hours, showing significant increase in fraction of Bcl11b-YFP⁺ cells on OP9-DL1 (40 hrs: χ²=6.1, d.f. = 1; p=0.013; 80 hrs: χ²=26.7, d.f. = 1; p=2.4×10⁻⁷). d) Time evolution for Bcl11b-YFP⁺ cells, obtained from Gaussian fits in (a). Smooth curves represent fits to the logistic equation f = A/[1+e^{4ν(t−τ)/A}]. Inset shows graphical depiction of parameters. Bars show initial Bcl11b-YFP activation rate (v), and maximal fraction of Bcl11b-YFP+ cells (A) from logistic fits, with 95% confidence intervals indicated. Results in (a,c,d,e) were obtained from one experiment, and results in (b) were obtained from a separate experiment, where cells were tracked to 36 hours. Bar graph data in (b,c) were derived from the indicated number of cells (n). Similar results were observed in three independent experiments.

Figure 6. Notch signaling increases Bcl11b activation probability in a dose-dependent manner

a–b) Flow cytometry analysis of BM Bcl11b-YFP⁻DN2 cells cultured on plates with surface-immobilized DL1 protein, and analyzed after 4 days. a) Levels of the dendritic cell marker CD11c (top), together with CD25 and Bcl11b-YFP levels (middle, bottom) in CD11c- cell populations. B) Percentage of CD11c+ dendritic cells (top), mean CD25 expression levels (center), and Bcl11b-YFP⁺ percentages (bottom) as a function of DL1 dosage. Curves represent best fit to hyperbolic function f(x) = x/(x + K_e) (bottom), or to a straight line (top, bottom), which was used due to lack of observable saturation in the concentration series. K_e = 0.28 µg/ml represents the DL1 dose required for half-maximal enhancement of Bcl11b activation. FSC - forward scatter. Results are representative of four independent experiments.

Figure 7. Gata3 and Tcf7 control initial Bcl11b activation

a)–b) Flow cytometry analysis of BM DN progenitors transduced with shRNA to Gata3 (shGata3) or Tcf7 (shTcf7), and cultured on OP9-DL1 monolayers with IL-7 and Flt3L for 5–6 days. a) CD44, CD25 and Bcl11b-YFP levels for the indicated cell populations. Cells transduced with non-targeting shRNA sequence (shRandom) are shown with dotted lines. b) Fraction of Bcl11b-YFP⁺ cells under different conditions. Solid green circles represent data from an alternate Tcf7 targeting construct (see materials and methods). Gata3 and Tcf7 knockdown significantly inhibits Bcl11b activation from ETPs, and in Bcl11b-YFP⁻DN2 progenitors to a lesser extent (unequal variance t-test on perturbed versus control conditions, 1-tailed; * p < 10⁻²; ** - p < 10⁻⁴), but shows little effect on Bcl11b expression after lineage commitment. Results in (a) are representative of two independent experiments, and data in (b) show means of the indicated number of replicates (n) from two independent experiments.

Figure 8. Runx1 controls Bcl11b expression amplitude

a)–b) Flow cytometry analysis of BM DN progenitors transduced with Runx constructs, and cultured on OP9-DL1 or OP9 monolayers for 4 days. CD44, CD25 and Bcl11b-YFP levels are shown for initial Bcl11b-YFP⁻DN2 cells (a) or Bcl11b-YFP⁺DN2 cells (b). shRunx1– Runx1 shRNA targeting sequence; RunxDN – dominant negative Runx inhibitor; Runx1 – full length Runx1; shRandom – non-targeting shRNA sequence; empty – empty vector. In histograms, dotted lines indicate Bcl11b-YFP levels in unperturbed cells. c) Mean Bcl11b-YFP levels under various Runx perturbation conditions. Altering Runx1 levels or activity significantly alters Bcl1b-YFP levels (unequal variance t-test, 2-tailed; * p < 0.05; ** - p < 0.01). d) Flow cytometry analysis of mature T cells activated with anti-TCRβ and anti-CD28, transduced with the indicated shRNA constructs, and cultured for 3 days. Data shows CD8 T cells (see also Fig. S8). Cells transduced with random shRNA (black solid), and non-fluorescent T cells (black dotted) are shown. Mean Bcl11b-YFP levels decrease significantly upon Runx1 knockdown (unequal variance t-test, 1-tailed; * p < 0.05). Results in (a,b) are representative of two (Runx, shRunx1) or three (RunxDN) independent experiments, and data in (c) represents data and means of the indicated number of replicates (n) from one experiment. Data in (d) represent means of two replicates from one experiment; results are representative of two independent experiments.