Evidence for coregulation of myocardial gene expression by MEF2 and NFAT in human heart failure

Abstract

Background: Pathological stresses induce heart failure in animal models through activation of multiple cardiac transcription factors (TFs) working cooperatively. However, interactions among TFs in human heart failure are less understood. Here, we use genomic data to examine the evidence that 5 candidate TF families coregulate gene expression in human heart failure.

Methods and results: RNA isolates from failing (n=86) and nonfailing (n=16) human hearts were hybridized with Affymetrix HU133A arrays. For each gene on the array, we determined conserved MEF2, NFAT, NKX , GATA , and FOX binding motifs within the -1-kb promoter region using human-murine sequence alignments and the TRANSFAC database. Across 9076 genes expressed in the heart, TF-binding motifs tended to cluster together in nonrandom patterns within promoters of specific genes (P values ranging from 10(-2) to 10(-21)), suggesting coregulation. We then modeled differential expression as a function of TF combinations present in promoter regions. Several combinations predicted increased odds of differential expression in the failing heart, with the highest odds ratios noted for genes containing both MEF2 and NFAT binding motifs together in the same promoter region (peak odds ratio, 3.47; P=0.005).

Conclusions: These findings provide genomic evidence for coregulation of myocardial gene expression by MEF2 and NFAT in human heart failure. In doing so, they extend the paradigm of combinatorial regulation of gene expression to the human heart and identify new target genes for mechanistic study. More broadly, we demonstrate how integrating diverse sources of genomic data yields novel insight into human cardiovascular disorders.

Figure 1

Overall design.

Figure 2

Observed versus expected frequency of TF combinations in promoters of N=9076 cardiac genes. The expected frequency for a specific TF combination was determined by assuming that the occurrence of each TF family in the combination was distributed randomly across all genes. A, Ratio of observed to expected frequency of genes with specific TF families as a function of the overall number of TF families. B, Observed (black) vs. expected (gray) frequencies of specific TF combinations in promoters of cardiac genes. Headings indicate the overall number of TF families. Bonferroni-corrected p-values for the exact binomial test comparing observed versus expected frequencies appear below each combination; P-values >.05 are indicated by a dashed line.

Figure 2

Observed versus expected frequency of TF combinations in promoters of N=9076 cardiac genes. The expected frequency for a specific TF combination was determined by assuming that the occurrence of each TF family in the combination was distributed randomly across all genes. A, Ratio of observed to expected frequency of genes with specific TF families as a function of the overall number of TF families. B, Observed (black) vs. expected (gray) frequencies of specific TF combinations in promoters of cardiac genes. Headings indicate the overall number of TF families. Bonferroni-corrected p-values for the exact binomial test comparing observed versus expected frequencies appear below each combination; P-values >.05 are indicated by a dashed line.

Figure 3

Comparison of odds ratios for differential expression using two different Tmax cutpoints across all TF combinations. TF combinations that include both MEF2 and NFAT (black) show the highest odds of differential expression.