CBP/p300 double null cells reveal effect of coactivator level and diversity on CREB transactivation

Abstract

It remains uncertain how the DNA sequence of mammalian genes influences the transcriptional response to extracellular signals. Here, we show that the number of CREB-binding sites (CREs) affects whether the related histone acetyltransferases (HATs) CREB-binding protein (CBP) and p300 are required for endogenous gene transcription. Fibroblasts with both CBP and p300 knocked-out had strongly attenuated histone H4 acetylation at CREB-target genes in response to cyclic-AMP, yet transcription was not uniformly inhibited. Interestingly, dependence on CBP/p300 was often different between reporter plasmids and endogenous genes. Transcription in the absence of CBP/p300 correlated with endogenous genes having more CREs, more bound CREB, and more CRTC2 (a non-HAT coactivator of CREB). Indeed, CRTC2 rescued cAMP-inducible expression for certain genes in CBP/p300 null cells and contributed to the CBP/p300-independent expression of other targets. Thus, endogenous genes with a greater local concentration and diversity of coactivators tend to have more resilient-inducible expression. This model suggests how gene expression patterns could be tuned by altering coactivator availability rather than by changing signal input or transcription factor levels.

Figure 1

In vitro deletion of CBP and p300 conditional alleles produces CBP/p300 double null MEFs. (A) Schematic of CBP/p300 double null (dKO) MEF derivation. LoxP sites in CBP^flox and p300^flox alleles flank an exon in the KIX domain (∼aa 600) and result in a frame shift mutation after recombination. eGFP expression from a recombination-dependent transgene shown. (B) Western blot of control (CBP^flox/flox; p300^flox/flox) and FACS purified GFP⁺ dKO MEFs. Antibodies to CBP and p300 protein were raised against both N- and C-terminal peptides. A non-specific band detected by the p300 antiserum was used as a loading control. Extracts were made 4 days post-Ad-Cre infection, immediately after FACS purification. (C) Indirect immunofluorescence in dKO MEFs showing loss of CBP/p300 proteins. MEFs were stained with a cocktail of antibodies against the N- and C-termini of CBP and p300. Closed yellow arrow marks example of deleted dKO cells. Open white arrows mark examples of non-deleted cells for internal controls. Closed white arrows mark cells null for only CBP or p300 showing partial loss of signal. DAPI stained nuclei. (D) CBP and p300 protein turnover in MEFs. Indirect immunofluorescence of CBP and p300 using antibodies against the C-termini of CBP and p300 in dKO MEFs 1–4 days after Ad-Cre infection. Wild-type cells treated with Ad-Cre are provided for IF signal comparison. DAPI stained nuclei. (E) Growth curve of WT and dKO MEFs. Percent of GFP⁺ cells expressing Cre recombinase after Ad-Cre infection was assessed by flow cytometry and applied to cell count. Two independent MEF isolates of each genotype used. (F) dKO MEFs in culture 16 days after floxed gene deletion. Closed arrows indicate GFP⁺ cells that are null for CBP and p300.

Figure 2

cAMP-inducible genes are unequally affected in CBP/p300 null MEFs. (A) Classic model of CREB-dependent transcription. Upon cAMP or calcium signalling, constitutively DNA-bound CREB is phosphorylated on serine 133 and CRTC translocates to the nucleus allowing the binding of CBP/p300 and CRTC to CREB and transcription of target genes. TFIID is constitutively bound to CREB. (B, C) Affymetrix expression analysis of wild type and CBP/p300 dKO MEFs treated 90 min with 10 μM forskolin+100 μM IBMX (FI) (mean of N=2 independent MEF isolates). (B) Operationally defined cAMP-induced genes; 347 cAMP-responsive probe sets defined by all present call in wild-type MEFs and an FI/EtOH expression signal ratio ⩾2.5 in both wild-type MEF isolates. (C) Operationally defined control genes; 204 non-cAMP-responsive probe sets defined by all present call in wild-type FI-treated samples and an FI/EtOH signal ratio of ±2% (non-induced) in both wild-type MEF samples. (D–H) qRT–PCR of wild type (WT) and dKO MEFs treated for 90 min with EtOH or FI. Relative gene expression values for each gene were normalized to Pgk1 mRNA expression and the lowest value for each gene was set to one. (D–G) Mean±s.e.m., N=4. (H) Heat map colours represent the z-score normalization of the mean expression values of each gene individually and are not strictly comparable between genes. Red indicates high expression values relative to blue. N=2–4.

Figure 3

cAMP-inducible histone H4 lysine acetylation at CREB-dependent promoters is CBP/p300 dependent, but not strictly required for gene expression. (A) Wild-type (WT) and dKO MEFs infected with GFP (control) or A-CREB (dominant negative that inhibits DNA binding for CREB family members) expressing retrovirus treated for 90 min with ethanol vehicle (EtOH) or 10 μM forskolin±100 μM IBMX (FI). Relative gene expression values for each gene were normalized to Pgk1 mRNA expression and the lowest value for each gene was set to one. Heat map colours represent the z-score normalization of the mean expression values of each gene individually and are not strictly comparable between genes. N=2. (B–F) ChIP at the CREs of 13 CREB-target gene promoters in WT MEFs treated for 30 min with EtOH or FI. Genes are arranged from most to least CBP/p300 dependent for expression based on data in Figure 1K. (B) ChIP signal for normal rabbit serum (NRS) or CBP/p300 was normalized to chromatin input signal. Heat map shows the mean log₂ normalized ChIP signals. N=2. (C–F) Histone H4 acetylated K5 (C), K8 (D), K12 (E) and K16 (F) ChIP was normalized to chromatin input signal, then to histone H4 ChIP signal from a non-discriminating H4 antibody. Heat maps show the mean normalized ChIP signals; N=2. In all heat maps (A–F), red indicates high values relative to blue. (G–J) WT and dKO MEFs, as well as dKO MEFs infected with CBP-HA retroviruses (WT, HAT mutant WY/AS or HAT mutant FPY/AAA) treated for 90 min with EtOH or FI. Relative expression values for each gene were normalized to Pgk1 mRNA expression and the lowest value for each gene was set to one. Mean ±s.e.m., N=2.

Figure 4

Increased transcription factor and coactivator recruitment correlates broadly with CBP/p300-independent gene expression. (A) Analysis of the promoters (−1 to +1 kb from TSS) of 172 genes with clearly annotated TSS induced at least 2.5-fold by 90 min 10 μM forskolin+100 μM IBMX (FI) treatment in wild-type (WT) MEFs, comparing relative dependence on CBP/p300 (log₂ of ratio of WT FI signal to CBP/p300 dKO FI signal) to number of CREs (TGACG). Expression data for unambiguously annotated genes in the Affymetrix experiment described in Figure 1 were used. (B) Pearson's correlation of the CREB ChIP signal in WT MEFs (treated 30 min with FI) at the CREs in 24 CREB-target gene promoters against the number of CREs in the region 600 base pairs centered on the primers used for ChIP analysis. (C–E) Pearson's correlation of CREB versus CBP/p300 (C), CRTC2 (D), and CBP/p300 versus CRTC2 (E) ChIP signals in WT MEFs treated 30 min with FI at the CREs in 24 CREB-target gene promoters. (F) Model showing that increased recruitment of multiple coactivator types at promoters with more bound transcription factor enhances transcriptional resilience. (G–I) Comparison of CREB (G), CBP/p300 (H), and CRTC2 (I) ChIP signals in WT MEFs treated 30 min with FI at the CREs in 24 CREB-target gene promoters with the relative dependence of these genes on CBP/p300 for their expression (expressed as the log₂ of ratio of WT FI signal to CBP/p300 dKO FI signal). All ChIP signals normalized to chromatin input signal (B–E and G–I). Pearson's correlation coefficients (r) and P-values indicated (where r=0 indicates no correlation and r=1 is a perfect positive correlation) (A–E and G–I).

Figure 5

CRTC2 overexpression rescues CBP/p300-dependent expression of some CREB-target genes. (A–D) ChIP of WT and dKO MEFs infected with control (GFP) or CRTC2 retrovirus using normal rabbit serum control (NRS), a mix of CBP and p300 antibodies or CRTC2 antibody, and primers at the CRE in CREB-target gene promoters. MEFs for ChIP were treated 30 min with 10 μM forskolin+100 μM IBMX (FI) or ethanol vehicle (EtOH). ChIP signal is normalized to chromatin input signal. Mean±s.e.m., N=2. (E–H) Gene expression measured by qRT–PCR of WT and dKO MEFs infected with control (GFP) or CRTC2 retrovirus and treated 90 min with EtOH or FI. Relative gene expression values for each gene were normalized to Pgk1 mRNA expression and the lowest value for each gene was set to one. Mean±s.e.m., N=2–3.

Figure 6

CRTC knockdown reduces CBP/p300-independent CREB-target gene expression in dKO MEFs. (A–H) Gene expression measured by qRT–PCR of WT and dKO MEFs 72 h after transfection with non-specific scramble siRNA or a mix of siRNAs targeting CRTC1 and CRTC2. In the presence of serum (which tends to dampen the fold response to cAMP), MEFs were treated for 90 min with 10 μM forskolin+100 μM IBMX (FI) or ethanol vehicle (EtOH) before RNA harvest. Mean±s.e.m., N=7. (I) Ratio of the average FI-induced gene expression of the eight genes in (A–H) in WT or dKO MEFs treated with Crtc1 plus Crtc2 siRNA relative to scramble siRNA. Mean±95% confidence interval. (J) IP western of CTRC2 in WT and dKO MEF whole cell extracts 3 days after transient transfection with scramble siRNA or siRNAs targeting CRTC1 and CRTC2. Western blot for actin in same extracts used for IP western in upper panel. Note increased CRTC2 protein in dKO MEFs compared with WT MEFs. (K, L) CRTC1 and CRTC2 gene expression measured by qRT–PCR of mRNA from WT and dKO MEFs transfected with a scramble siRNA or a mixture of siRNAs against CRTC1 and CRTC2 (CRTC siRNA) and treated 90 min with FI or ethanol vehicle (EtOH) (N=7, mean±s.e.m.).