Functional Definition of Thyroid Hormone Response Elements Based on a Synthetic STARR-seq Screen

Abstract

When bound to thyroid hormone, the nuclear receptor TRα1 activates the transcription of a number of genes in many cell types. It mainly acts by binding DNA as a heterodimer with retinoid X receptors at specific response elements related to the DR4 consensus sequence. However, the number of DR4-like elements in the genome exceed by far the number of occupied sites, indicating that minor variations in nucleotides composition deeply influence the DNA-binding capacity and transactivation activity of TRα1. An improved protocol of synthetic self-transcribing active regulatory region sequencing was used to quantitatively assess the transcriptional activity of thousands of synthetic sites in parallel. This functional screen highlights a strong correlation between the affinity of the heterodimers for DNA and their capacity to mediate the thyroid hormone response.

Keywords: hormone response elements; nuclear receptors; thyroid hormone.

Figure 1.

Library description. (A) Logo of consensus DR4-like sequence discovered by de novo motif search (25) from a TRα1 ChIP-seq dataset obtained from mouse striatum neurons (11). According to structural data (15) the 5′ half-site (5′RGGTCA3′) is occupied by RXR. The 2 3′ nucleotides of the 4 nucleotides spacer and the 3′ half-site (5′NNRGGNCA3′) contact TRα1. (B) The expression vector is transcribed in eukaryotic cells from a cryptic minimal promoter present in the Ori sequence, spliced, and contains the cloned amplicon. Sequences flanking the DR4 element with degenerated nucleotides are flanked by sequences present in the Hairless gene promoter. The small arrows indicate the position of the primer used for cDNA amplification. (C) Sanger sequencing of the input library (left) indicates the presence of a mixture of different nucleotides at the expected positions. Sequencing of cDNA prepared from transfected cells (right) confirm the heterogeneity of the library and suggests that T3 has an influence on the respective abundance of different mRNA/DR4-like elements in the amplified cDNA.

Figure 2.

Features of the DR4 response element favoring T3 response. (A) Frequency of specific dinucleotides in the downstream half-site contacted by TRα1 (positions 14 and 15). The table indicates the enrichment in cDNA library prepared from T3 treated cells compared with the input library. (B) Frequency of specific tetranucleotides in the spacer (positions 7 to 10). The table indicates the enrichment in cDNA library prepared from T3-treated cells compared with the input library for the 5 most favorable spacer tetranucleotides. (C) Global analysis of all tested DR4 (Deseq2). Each gray dot represents a specific sequence. Black triangles are for the 455 sequences significantly enriched in the cDNA library prepared from T3-treated cells compared with the input library (adjusted P < 0.05). The 5 most efficient DR4 sequences are indicated and listed in the table.

Figure 3.

Individual tests of 9 DR4 sequences in transient expression assays. DR4-1 is the most efficient of the tested DR4, according to Fig. 2C. Comparisons between DR4-1, -2, -3, and -4 highlight the most favorable nucleotide in the 3′ half-site occupied by TRα1. Comparisons with DR4, -5, -6, -7, and -8 demonstrate that spacer nucleotides have an important influence on transactivation and that a T at the third position of the spacer (position 9) is more favorable to transactivation. DR4-9 was predicted to be inactive by STARR-seq analysis, which is confirmed in this assay. Note that the negative influence exerted by unliganded TRα1, and that induction rate is maximal for DR-7. Two to 4 repeats of DR4-6 were previously used in reporter constructs to provide maximal transactivation capacity (18).