Genetic interaction between mutations in c-Myb and the KIX domains of CBP and p300 affects multiple blood cell lineages and influences both gene activation and repression

Abstract

Adult blood cell production or definitive hematopoiesis requires the transcription factor c-Myb. The closely related KAT3 histone acetyltransferases CBP (CREBBP) and p300 (EP300) bind c-Myb through their KIX domains and mice homozygous for a p300 KIX domain mutation exhibit multiple blood defects. Perplexingly, mice homozygous for the same KIX domain mutation in CBP have normal blood. Here we test the hypothesis that the CBP KIX domain contributes subordinately to hematopoiesis via a genetic interaction with c-Myb. We assessed hematopoiesis in mice bearing compound mutations of c-Myb and/or the KIX domains of CBP and p300, and measured the effect of KIX domain mutations on c-Myb-dependent gene expression. We found that in the context of a p300 KIX mutation, the CBP KIX domain mutation affects platelets, B cells, T cells, and red cells. Gene interaction (epistasis) analysis provides mechanistic evidence that blood defects in KIX mutant mice are consistent with reduced c-Myb and KIX interaction. Lastly, we demonstrated that the CBP and p300 KIX domains contribute to both c-Myb-dependent gene activation and repression. Together these results suggest that the KIX domains of CBP, and especially p300, are principal mediators of c-Myb-dependent gene activation and repression that is required for definitive hematopoiesis.

Figure 1. The KIX domains of both CBP and p300 contribute to c-Myb transactivation function in a dose dependent manner.

(A) Schematic showing approximate locations of CBP and p300 domains [91]. Nuclear receptor interaction domain (NRID), the CREB-binding domain (KIX), Cys/His-rich region-1 (CH1 [57] or transcriptional-adaptor zinc-finger-1, TAZ1), bromodomain (Br), plant homeodomain (PHD), HAT enzymatic region, zinc-binding domain (ZZ), TAZ2 (aka CH3 [92]), and the nuclear receptor binding domain (NCBD). (B) Alignment of the KIX domains of mouse and human p300 and CBP. Amino acids mutated in the KIX domain mutants are indicated with arrows. (C) Pulldown assay with the c-Myb transactivation domain (aa186-325) fused to GST using extracts from wild type (WT) and CBP ^KIX/KIX mouse embryonic fibroblasts. Buffer C is extraction buffer without protein. (D) Transient transfection assay showing Gal-c-Myb 186-325 transactivation function is reduced in mouse embryonic fibroblasts having the indicated number of CBP and p300 mutant KIX alleles (one (p300^+/KIX or CBP^+/KIX), two (CBP^KIX/KIX, p300^KIX/KIX or p300^+/KIX;CBP^+/KIX) or three (p300^+/KIX;CBP^KIX/KIX or p300^KIX/KIX;CBP ^+/KIX) KIX); wild type MEFs indicated; one MEF isolate of each genotype except wild type (N=2); (mean ± SEM).

Figure 2. The KIX domains of CBP and p300 contribute to adult blood cell production in the fetus, albeit nonequivalently.

(A-D) e15.5 fetal liver sections, H & E staining, 400x magnification. (E) Flow cytometry on e15.5 fetal liver cells showing an increased CD41⁺ megakaryocyte population in p300^KIX/KIX and p300^KIX/KIX;CBP ^+/KIX mice. (F) e15.5 embryos demonstrating anemia-associated pallor in p300^KIX/KIX, p300^KIX/KIX;CBP ^+/KIX and c-Myb^-/- embryos. (G,H) Cell counts for e15.5 fetal livers. For both the KIX and Myb mutant panels, the analysis was a one-way ANOVA (p<0.0001) with Tukey post test between genotype pairs. Non significant pairings are not marked and significant pairings are indicated * p<0.05, ** p<0.01 and *** p<0.001. For the KIX panel, N=4-11 for each genotype, for the Myb panel N=2 (Myb KO) to 5. (I) CD25/CD44 flow cytometry on e15.5 fetal thymocytes gated first for CD4^-CD8^- (double negative, DN) subset. Embryos are C57BL/6J x 129Sv F2 background except for c-Myb^+/- and c-Myb^-/- embryos, which were C57BL/6J. DN stages shown in wild type.

Figure 3. Triple-KIX p300^+/KIX; CBP ^KIX/KIX adult mice have defective hematopoiesis affecting multiple lineages.

Peripheral blood counts from 1-4 month old C57BL/6J x 129Sv F1 wild type (WT) and KIX mutant mice (N=10-22). Counts from automated Hemavet complete blood count. For B cells and T cells, total lymphocyte counts from complete blood count were parsed using flow cytometry to determine the percentage of lymphocytes that were positive for B220 (B cells), CD3 (T cells), CD4 and CD8 (T cell subsets). Asterisks indicate significant p value by pairwise Tukey post test following one way ANOVA (* p<0.05, ** p<0.01, *** p<0.001).

Figure 4. Combined KIX domain and c-Myb haploinsufficiency synergistically affects multiple blood lineages.

Peripheral blood counts from 1-3 month old C57BL/6J x 129Sv F1 and C57BL/6J background mice. (A-G) Counts from automated Hemavet complete blood count. For B cells and T cells, total lymphocyte counts from complete blood count were parsed using flow cytometry to determine the percentage of lymphocytes that were positive for B220 (B cells), CD3 (T cells), CD4 and CD8 (T cell subsets). Asterisks indicate significant p value by pairwise Tukey post test following one way ANOVA (N=13-19 per genotype, * p<0.05, ** p<0.01, *** p<0.001). (H) Epistasis values and 95% confidence intervals calculated using five independent experiments with wild type, c-Myb^+/-, p300^+/KIX;CBP ^+/KIX and triple-het p300^+/KIX;CBP^+/KIX;c-Myb^+/- mice represented in each. Zero indicates no epistasis or the lack of synergistic interaction between mutations [i.e. equality of the multiplicative phenotype (or additivity of log-transformed phenotype in this case) of the extreme (wild type and p300^+/KIX;CBP^+/KIX;c-Myb^+/-) and intermediate genotypes (c-Myb^+/- and p300^+/KIX;CBP ^+/KIX)]. A positive epistasis value indicates a greater than multiplicative increased phenotype for the extreme genotype, a negative epistasis value indicates a greater than multiplicative decreased (aggravated) phenotype for the extreme genotype.

Figure 5. Mutation of the KIX domains of CBP and p300 affects expression of both c-Myb activated and repressed genes.

(A,B) Affymetrix gene array of wild type (WT) and p300^KIX/KIX;CBP ^+/KIX primary mouse embryonic fibroblasts (MEFs) transduced with MSCV c-Myb IRES GFP (c-Myb) retrovirus. Data shown represents average of two biological replicates for each genotype. (A) probe sets shown were scored all present in both WT + c-Myb MEFs. 509 probe sets (marked in black) were induced at least twofold by c-Myb expression in both WT MEFs. (B) probe sets shown were scored all present in WT and p300^KIX/KIX;CBP ^+/KIX + MSCV IRES GFP control retrovirus MEFs. 148 probe sets (marked in black) were repressed at least 50% by c-Myb expression in both WT MEFs. Axes scales are log₂.

Figure 6. Both c-Myb activated and repressed endogenous genes are sensitive to mutation of the KIX domains of CBP and p300 in thymocytes.

(A,B) Affymetrix gene array of wild type (WT) and triple-KIX p300^+/KIX ; CBP ^KIX/KIX CD4⁺CD8⁺ double positive (DP) thymocytes. Average signal; N=4 mice for each genotype. Probe sets marked in black were identified from data presented in Yuan et al [70]. and indicate probe sets in which c-Myb^flox/flox;Cd4-Cre;Tcrα ^-/- DP thymocytes were at least twofold higher (A) or at least 50% lower (B) than control cells. For our data sets (black and blue icons), no probes are shown with a signal below 3.4 (twice background). For the Myb-dependent gene set from Yuan et al. used here, probes selected had P<0.05 between Myb null and control DP thymocytes, and none were used where all signals were below 6. Axes scales are log₂.

Figure 7. KIX mutation-sensitive genes in DP thymocytes are enriched for genes that are also activated and repressed in c-Myb null DP thymocytes.

Gene set enrichment analysis (GSEA [75]) using Affymetrix gene expression data from wild type (WT) and p300^+/KIX; CBP ^KIX/KIX (triple-KIX) CD4⁺CD8⁺ double positive (DP) thymocytes found significant (FDR q = 0.0, FWER p = 0.0) for enrichment of c-Myb activated (A,B) and repressed (C,D) genes. The two GSEA gene sets shown here were derived from gene expression data for c-Myb ^+/flox; Cd4-Cre; Tcrα^-/- (control) and c-Myb ^flox/flox; Cd4-Cre; Tcrα^-/- (c-Myb null) DP thymocytes [70], which includes c-Myb activated (at least two fold higher in control) and repressed (at least 50% lower in control) genes. In enrichment plots (A, C), genes are ranked by signal/noise ratio according to their differential expression between WT and Triple-KIX DP thymocytes. Genes in the c-Myb gene sets are marked with vertical bars, and the enrichment score is shown in green. Relative expression of the c-Myb activated (B) and repressed (D) gene sets in WT and Triple-KIX DP thymocytes (N=4 mice each) are shown in order of their signal/noise ratio rank.

Figure 8. Il2ra (CD25) is abnormally expressed on CD4⁺CD8⁺ thymocytes from mice with KIX and c-Myb complex insufficiencies.

(A) qRT-PCR for Il2ra mRNA in CD4⁺CD8⁺ double positive (DP) thymocytes from wild type (WT) and triple-KIX p300^+/KIX;CBP ^KIX/KIX mice. (B-E) CD25 protein expression on DP thymocytes from wild type (B,D), triple-KIX p300^+/KIX;CBP^KIX/KIX (C) and combined haploinsufficient triple-het p300^+/KIX;CBP^+/KIX;c-Myb^+/- (E) mice. 5 week old C57BL/6J x 129Sv F1 (B,C) and 8 week old C57BL/6J (D,E) mice.

Figure 9. c-Myb dependent gene set identified in c-Myb null Lin^- Sca-1⁺ c-kit⁺ (LSK) cells is enriched in KIX mutant DP thymocytes, but displays a largely reversed expression profile from expected.

Gene set enrichment analysis (GSEA) using Affymetrix gene expression data from wild type (WT) and p300^+/KIX; CBP ^KIX/KIX (Tri-KIX) CD4⁺CD8⁺ double positive (DP) thymocytes found significant (FDR q = 0.00079, FWER p = 0.0030) enrichment of a set of genes dependent on c-Myb in LSK cells as defined by [71]. (A) In the enrichment plot, genes are ranked by signal/noise ratio according to their differential expression between WT and Tri-KIX DP thymocytes. Genes in the LSK gene set are marked with vertical bars, and the enrichment score is shown in green. (B) Relative expression of the LSK gene set in WT and Tri-KIX DP thymocytes (N=4 mice each) are shown in order of their signal/noise ratio rank.