Expression profiles frame the promoter specificity dilemma of the ETS family of transcription factors

Abstract

Sequence-specific DNA binding proteins that function as transcription factors are frequently encoded by gene families. Such proteins display highly conserved DNA binding properties, yet are expected to retain promoter selectivity. In this report we investigate this problem using the ets gene family, a group of metazoan genes whose members regulate cell growth and differentiation and are mutated in human cancers. We tested whether the level of mRNA can serve as a specificity determinant. The mRNA levels of the 27 paralogous human ets genes were measured in 23 tissues and cell lines. Real-time RT-PCR provided accurate measurement of absolute mRNA levels for each gene down to one copy per cell. Surprisingly, at least 16 paralogs were expressed in each cell sample and over half were expressed ubiquitously. Tissues and complementary cell lines showed similar expression patterns, indicating that tissue complexity was not a limitation. There was no unique, highly expressed gene for each cell type. Instead, one of only eight ets genes showed the highest expression in all samples. DNA binding studies illustrate both overlapping and unique specificities for ubiquitous ETS proteins. These findings establish the parameters of the promoter specificity dilemma within the ets family of transcription factors.

Figure 1

Absolute values for mRNA levels are accurate to one copy per cell. The expression of ETS2, ER81, E1AF, ESE2 and SPIB in 23 cell samples was measured by real-time RT–PCR with gene specific primers (Table 1). Values were converted to mRNA copies per cell by the use of a standard curve and 18S rRNA as an internal control. Each cDNA measurement was repeated with an independent primer set. The fold difference (value 1/value 2) between the two measurements was 2-fold or less in the gray area. Of 61 measurements equal to or greater than one copy per cell, 59 (97%) showed <2-fold error. This error was not significantly different for each individual gene and can therefore be considered gene independent.

Figure 2

Expression profiles for ets gene family demonstrate extensive co-expression. The expression of 27 human ets genes was measured by real-time RT–PCR with gene specific primers (Table 1). Horizontal lines separate ets genes into subgroups that are defined by similarity in the ETS domain. The mRNA copy number per cell was estimated as mRNA molecules per 2 × 10⁶ 18S rRNA molecules in the same sample. Values <1, indicated with an asterisk, could not be measured accurately (Figure 1). Each column represents values from a single RNA sample. Values for ESE2, ETS2, SPIB, E1AF and ER81 are the mean of mRNA level of the same cDNA sample with two independent primer sets. Since simple repetition gave much lower error (see Materials and Methods) than that inherent in the assay (Figure 1) such measurements were not deemed valuable and the error for all values should be assumed to be ∼2-fold. Values in bold indicate the most highly expressed ets gene in a cell sample.

Figure 3

Ets family profile in tissues and matching cell lines are similar. The mRNA copy number per cell for all 27 human ets genes is compared between a tissue sample and a cell line representing a major cell type in that tissue (data from Figure 2). The off-scale mRNA values of ESE1 in colon and ESE3 in prostate are 449 and 669, respectively.

Figure 4

ETS protein promoter occupancy varies in specificity. (A) Chromatin immunoprecipitations were performed with an antibody that recognizes both ETS1 and ETS2 (ETS1/2), or antibodies specific to ETS1, ETS2, or ELK1, or pre-immune rabbit IgG in cell lines indicated. Immunoprecipitated DNA was analyzed by PCR specific to the promoter of EGR1, or CDC2L2, or a 3′ region of the albumin gene. (B) Chromatin immunoprecipitations were performed as in (A). Real-time PCR measured enrichment of a negative control region of the albumin gene and the promoter of CDC2L2 by each ETS antisera. Measurements are expressed as a fold enrichment over the IgG control, therefore, lack of a specific enrichment is represented by a value of one. Values are an average (±standard error) from two independent experiments, except for ETS2 in HCT116, which was performed only once.