T-cell differentiation is blocked in a CBFβ dosage-dependent manner. (A) Proliferation of Lin⁻ (Ter119, CD19, Gr-1, F4/80) 17.5 dpc FL cells cocultured on OP9-DL1 in the presence of Flt3L and IL-7, counted at weekly intervals. Each data point is averaged from at least 9 fetal livers. Error bars in all panels represent SEM. (B) Bar graph depicting the fold expansion of Lin⁻ FL cells from an allelic Cbfb dosage reduction series, after 14 days of OP9-DL1 coculture. The percentage of CBFβ protein relative to wild-type levels is indicated below the bars.¹³ The number of independent animals analyzed was: Cbfb^+/+ = 13; Cbfb^+/− = 7; Cbfb^rss/rss = 5; Cbfb^rss/− = 10. *indicates significant (P < .05) difference relative to Cbfb^+/+ animals (unpaired, 2-tailed Student t test). (C) CD4 and CD8 expression after 14 days of coculture on OP9-DL1. (D) Mean (± SEM) percentages of CD4⁺CD8⁺ (DP) cells. Cbfb^+/+ = 17; Cbfb^+/− = 7; Cbfb^rss/rss = 5; Cbfb^rss/− = 16. *Significant (P < .05) difference relative to Cbfb^+/+ animals. **indicates significant (P < .05) difference relative to Cbfb^rss/rss animals. (E) Cell surface TCRγ expression in the bulk cocultured cells. (F) Mean (± SEM) percentages of TCRγ⁺ cells after 14 days of culture. Cbfb^+/+ = 17; Cbfb^+/− = 7; Cbfb^rss/rss = 5; Cbfb^rss/− = 16. (G) Intracellular TCRβ staining in DN cells (DN = CD8⁻Gr-1⁻ because CD4 expression is inappropriately unregulated in Cbfb^rss/− DN cells^13,72) after 6 and 9 days of OP9-DL1 culture. Cbfb^+/+ = 4; Cbfb^rss/− = 3.

JAK/STAT and c-kit signaling in Cbfb^rss/− T cells. (A) Expression of IL-7Rα (CD127) and γ_c (IL2RG/CD132) on the DN2 thymocytes from Figure 2C (17.5 dpc). Data are representative of 4 experiments. (B) DN2 cells were sorted from OP9-DL1 cultures (culture conditions and staining as described in Figure 2A). The purity of postsort populations was more than 95%. Expression of individual genes was normalized to Hprt. Values are averaged from triplicate samples from 3 independent experiments. *indicates significant difference (P≤ .05). (C) pStat5 levels in 17.5 dpc DN2 thymocytes on ex vivo stimulation with 0, 1, and 5 ng/mL IL-7 for 20 minutes. Shown is the MFI of the difference between pStat5 under IL-7–stimulated and nonstimulated conditions. The data are averaged from 5 Cbfb^+/+ and 5 Cbfb^rss/− fetuses. Differences are significant at P < .001. (D) Rescue of T-cell development from Cbfb^rss/−Lin⁻ FL cells with CBFβ, but not wild-type or constitutively active Stat5a (S711F; Stat5a^F), or MigR1 alone (−) (n = 5). Cells were harvested after 2 weeks. The expansion of Cbfb^rss/− cells (left graph) is calculated by dividing the number of GFP⁺ cells on day 14 by that on day 7 of culture. Only CBFβ expression significantly (*P≤ .001) increased cell numbers compared with MigR1-transduced Cbfb^rss/− cells. The right-hand graph shows the percentage of GFP⁺ cells expressing CD4 and CD8. (E) Expression of cell surface c-kit on 17.5 dpc DN1 cells. Bar graph on left is the percentage of c-kit⁺ DN1 (Lin⁻ as in Figure 2B; CD45⁺CD44⁺CD25⁻) thymocytes, and on right is the MFI of c-kit staining on c-kit⁺ DN1 thymocytes. The data are averaged from 4 Cbfb^+/+ and 4 Cbfb^rss/− fetuses. (F) Percentage of DN1 (CD44⁺CD25⁻CD45⁺ Lin⁻) cells expressing surface c-kit and MFI of c-kit staining on c-kit⁺ DN1 cells after 6 days of OP9-DL1 culture. The data are averaged from 4 Cbfb^+/+ and 3 Cbfb^rss/− fetuses.

Inhibition of Notch signaling does not affect Runx1, Runx3, or Cbfb expression. (A) Flow cytometric analysis of Lin⁻ FL cells cultured on OP9-DL1 in the absence and presence of the indicated concentrations of GSI for 7 days. CD8⁻CD19⁻Gr-1⁻GFP⁻ cells were analyzed for CD44 and CD25 expression (n_experiments = 8). (B) Gene expression profile of DN1 cells sorted from day 7 OP9-DL1 cocultures treated with dimethyl sulfoxide (□), 1.0 μM GSI (), or 3.0 μM GSI (■). Cell sorting, RNA preparation, and real-time PCR were performed as described for Figure 4A. Taqman probes were used for the quantification of Runx1, Runx3, Cbfb, and Hprt, and SYBR green was used for the remainder of the genes. Expression of each gene was quantified in comparison to a standard curve prepared with dilutions of spleen cDNA. The expression of individual genes is displayed relative to Hprt. Data are averaged from 4 independent experiments. Error bars represent SEM. *indicates significant difference (P≤ .05) between GSI-treated and untreated cells. (C) Runx1 expression in lymphoid progenitors (c-kit⁺CD27⁺CD25⁻) isolated from day 3 OP9-DL1 cultures in the absence and presence of 3 μM GSI (averaged from triplicate samples from 3 independent experiments). The increase in Runx1 expression in GSI-treated cultures was significant at P < .01. (D) Ectopic expression of Runx1 in Cbfb^+/+ FL cells cultured on OP9-DL1 in the absence and presence of GSI at 3 μM. Analysis was performed as in panel A.

Genetic interactions between the CBFs and Notch signaling in the specification of HSCs, T cells, and NK cells in the fetus. Notch receptors and ligands are expressed on the aortic endothelial cells in the AGM region that give rise to HSCs.²⁰ Notch signaling is required for Runx1 expression in endothelial cells and for the formation of HSCs^19,20,23,24; thus, Notch is genetically upstream of Runx1 in HSC formation. — represent molecular interactions; ---- represent cell migrations. PB indicates peripheral blood. HSCs and progenitors are released into the circulation from their sites of formation and colonize the fetal liver. We speculate that Runx1 and/or Runx3 plus CBFβ are required to generate NKP from either a bipotent T/NK or another progenitor in the fetus, and perhaps in T/NK progenitors themselves. CBFs are also required for the NKP to iNK transition, based on data from the adult. Circulating progenitors expressing all 3 CBF complexes colonize the thymus where they encounter high levels of Notch ligands. CBFs confer on these progenitors the ability to respond to Notch signaling, which results in T-cell specification and progression to the ETP/DN2 stage accompanied by the expression of a suite of T cell–specific genes. Not shown are NKT cells, which differentiate from DP T cells in a Runx1-dependent manner.

T-cell specification fails to occur in Cbfb^rss/− cells. (A) Lin⁻ FL cells cocultured on OP9-DL1 in the presence of Flt3L and IL-7 for 7 days. CD8⁻Gr-1⁻GFP⁻ cells (GFP gating was used to eliminate OP9-DL1) were analyzed for CD44 and CD25 expression. (B) Percentages of DN1, DN2, and DN3 cells (from panel A) averaged from 12 independent experiments. Error bars represent SEM. *indicates significant difference (P ≤ .01) between the percentages of Cbfb^+/+ and Cbfb^rss/− cells. (C) CD45⁺ Lin⁻ (Lin = CD8, CD3, TCRβ, TCRγ, CD11c, B220, Mac1, NK1.1, Ter119) 17.5 dpc fetal thymocytes analyzed for CD44 and CD25 expression. (D) Data from 6 animals of each genotype (from panel C). *indicates significant difference (P ≤ .03) between the percentages of Cbfb^+/+ and Cbfb^rss/− cells. (E) Expression of cell surface Thy1.2 on 17.5 dpc DN1 (Lin⁻CD45⁺CD44⁺CD25⁻) thymocytes, and on DN1 cells (CD8⁻Gr-1⁻CD45⁺CD44⁺CD25⁻) from day 6 OP9-DL1 cultures. The thymocyte data were averaged from 4 Cbfb^+/+ and 4 Cbfb^rss/− fetuses, and the culture data were averaged from 4 Cbfb^+/+ and 3 Cbfb^rss/− fetuses. *indicates significant difference (P = .01) between Cbfb^+/+ and Cbfb^rss/− samples. (F) Percentage of 17.5 dpc DN thymocytes (CD45⁺ Lin⁻, Lin = CD8, CD3, CD11c, B220, Mac-1, Ter119) expressing intracellular TCRβ (iTCRβ) and iTCRγ. Data are averaged from 3 Cbfb^+/+ and 3 Cbfb^rss/− fetuses. (G) Percentage of Thy1.2⁺ DN2 cells after 7 days of OP9-DL1 culture. (H) Cd3e and Cd3g expression by qRT-PCR in DN2 cells sorted from day 7 OP9-DL1 cultures (culture conditions and staining as described in panel A). The purity of postsort populations was more than 95%. Expression of individual genes was normalized to Hprt. Data were derived from triplicate amplifications from 3 independent samples. *indicates significant difference (P≤ .01) between Cbfb^+/+ versus Cbfb^rss/− values. (I) Lin⁻ FL cells cultured on OP9-DL1 cells, harvested at indicated time points. DN1 cells (Lin⁻CD45⁺CD44⁺CD25⁻) were isolated by cell sorting (purity > 95%). The expression of individual genes was normalized to Hprt (note log scale for Tcf7) and displayed as relative to day 0 values of Cbfb^+/+ DN1 cells. Values are averaged from triplicate samples isolated from 3 independent experiments. *indicates significant difference (P ≤ .05) between Cbfb^+/+ versus Cbfb^rss/− values. (J) Rescue of T-cell development with CBFβ in 17.5 dpc Cbfb^rss/− FL cells. Lin⁻ FL cells (E17.5) were transduced with the indicated retroviruses and cocultured on OP9-DL1. Cells were harvested after 2 weeks and GFP⁺CD45⁺ cells analyzed for CD4 and CD8 expression (top panels) and TCRγ expression (middle panels). The bottom 2 plots are DN cells analyzed after 1 week of coculture. Gated GFP⁺Lin⁻CD45⁺ cells were analyzed for expression of CD44 and CD25 (n_experiments = 11). (K) Western blot showing CBFβ (p22) protein levels resulting from retroviral expression relative to endogenous protein levels, in whole cell extracts prepared from GFP⁺CD45⁺ cells purified from the OP9-DL1 cultures of Lin⁻ FL cells (CD45⁺ cell purity ≥ 99.9%). The blot was probed with a monoclonal antibody to CBFβ. The 2 endogenous CBFβ isoforms generated as a result of alternative splicing (p21.5 and p22) are both visible on this gel. Samples were normalized for actin expression, and the relative amounts of CBFβ determined from a dilution series (not shown). CBFβ levels in retrovirally transduced cells were 5-fold higher than endogenous levels. Lane 1, Cbfb^rss/− + MigR1; lane 2, Cbfb^rss/− + CBFβ; lane 3, Cbfb^+/+ + MigR1; lane 4, Cbfb^+/+ + CBFβ. (L) Inefficient rescue of T-cell development on restoring CBFβ expression in 17.5 dpc Cbfb^rss/− thymocytes. Thymocytes were transduced with the indicated retroviruses and cocultured for 7 days on OP9-DL1. Gated GFP⁺CD45⁺ cells were analyzed for CD4 and CD8 expression in the plots below. GFP⁺CD45⁻ cells are predominantly OP9-DL1 cells.¹⁶ (M) Cell-cycle status of CD45⁺ Lin⁻ (Lin = CD8, TCRβ, TCRγ, CD11b, Ter119, B220) 17.5 dpc fetal thymocytes. (N) Summary of cell-cycle data from 5 individual samples per genotype. The differences between Cbfb^+/+ and Cbfb^rss/− cells in G₀/G₁ and G₂/M were significant at P≤ .01.

Notch signaling is active in Cbfb^rss/− cells and NK-cell development is defective. (A) Lin⁻ FL cells cultured on OP9-DL1 cells, harvested at indicated time points, and stained with antibodies as described in Figure 2A. DN1 cells (Lin⁻CD45⁺CD44⁺CD25⁻) were isolated by cell sorting (purity > 95%). The expression of individual genes was normalized to Hprt (note log scale for Dxt1) and displayed as relative to day 0 Cbfb^+/+ DN1 cells. Values are averaged from 9 samples (triplicate samples from 3 independent experiments). *indicates significant difference (P≤ .05) between Cbfb^+/+ versus Cbfb^rss/− values. (B) Flow cytometric analysis of Lin⁻ FL cells cultured on OP9-DL1 (+ Flt3L, IL-7, IL-6, IL-15) in the absence and presence of the gamma secretase inhibitor (GSI) inhibitor X for 7 days. GFP⁻CD45⁺ cells were analyzed for expression of NK1.1 (NK cells) and CD19 (B cells). n_experiments = 7. (C) Lin⁻Cbfb^+/+ 14.5 dpc FL cells transduced with either MigR1 or a retrovirus expressing the Notch1 intracellular domain (ICN) and cultured for 7 days on OP9 stromal cells in the presence of Flt3L and IL-7. CD8⁻CD19⁻Gr-1⁻ cells were gated for GFP expression in the left-hand panels, and GFP⁺ cells analyzed for CD44 and CD25 expression in the right-hand panels. (D) Lin⁻Cbfb^rss/− FL cells transduced with MigR1 or ICN, and analyzed as in panel C.

CBFβ is required at an early stage of NK-cell development in vivo. (A) Representative scatter plots of Lin⁻ 15.5 dpc FL cells grown on OP9 cells in the presence of Flt3L, IL-7, and IL-15. Gr-1⁻CD19⁻ cells were analyzed for CD122 and NK1.1 expression. (B) Percentage and absolute number of CD122⁺ cells (± SD) harvested from OP9 cultures established from 3 Cbfb^+/+ and 4 Cbfb^rss/− fetuses (n_experiments = 3). (C) NK cells in 15.5 dpc fetal thymi. (D) Total number of thymic NK1.1⁺ and NK1.1⁺CD122⁺ cells averaged from 3 Cbfb^+/+ and 3 Cbfb^rss/− fetuses (n_experiments = 3). (E-G) Lethally irradiated CD45.1 × CD45.2 F1 recipients were reconstituted with wild-type CD45.1⁺ BM and CD45.2⁺ FL cells from Cbfb^+/+, Cbfb^rss/rss, or Cbfb^rss/− fetuses (expressing 100%, 30%, and 15% of normal CBFβ levels, respectively). (E) Bone marrow of recipient mice analyzed 10 months after reconstitution. The data are from gated FL-derived (CD45.2⁺) progenitors. A representative example is shown in each group (Cbfb^+/+, n = 4; Cbfb^rss/rss, n = 4; Cbfb^rss/−, n = 5). Bone marrow from a 10-week-old C57BL/6 (B6) mouse is shown as control. CD19⁻CD3⁻TCRβ⁻NK1.1⁺DX5⁺ cells are mature NK cells, whereas immature NK lineage cells (iNK) have a NK1.1⁺Dx5⁻ phenotype. (F) The spleen of recipient mice analyzed in a similar way, illustrating the absence of immature and mature NK cells among cells derived from Cbfb^rss/− progenitors. (G) Primitive NK lineage committed progenitors (NKP) (CD122/IL2Rβ⁺NKG2D⁺NK1.1⁻Dx5⁻LinT⁻) (LinT = CD3CD4CD8αTCRβTCRγ) are preserved in the spleen of recipient mice. NKP express the IL-2/IL-15Rβ chain and have an NK1.1⁻Dx5⁻CD19⁻CD3⁻CD4⁻CD8a⁻TCRβ⁻TCRγ⁻ phenotype. A fraction of these cells expresses NKG2D.³ (H) Expression Runx1, Runx3, and Cbfb by qRT-PCR in LSK (Lin = CD4, CD8, TCRβ, TCRγ, DX5, CD19, Mac-1, Ter119), iNK (CD4⁻CD8⁻TCRβ⁻TCRγ⁻CD19⁻CD49b⁻NK1.1⁺CD122⁺) and mature NK (CD4⁻CD8⁻TCRβ⁻TCRγ⁻CD19⁻CD49b⁺NK1.1⁺CD122⁺) cells sorted from BM of 8- to 12-week-old wild-type mice. The purity of post-sort populations was more than 98%. Expression of individual genes was normalized to Hprt. * Significant difference (P≤ .01) from LSK cell values.