Evolution of the Antisense Overlap between Genes for Thyroid Hormone Receptor and Rev-erbα and Characterization of an Exonic G-Rich Element That Regulates Splicing of TRα2 mRNA

Abstract

The α-thyroid hormone receptor gene (TRα) codes for two functionally distinct proteins: TRα1, the α-thyroid hormone receptor; and TRα2, a non-hormone-binding variant. The final exon of TRα2 mRNA overlaps the 3' end of Rev-erbα mRNA, which encodes another nuclear receptor on the opposite strand of DNA. To understand the evolution of this antisense overlap, we sequenced these genes and mRNAs in the platypus Orthorhynchus anatinus. Despite its strong homology with other mammals, the platypus TRα/Rev-erbα locus lacks elements essential for expression of TRα2. Comparative analysis suggests that alternative splicing of TRα2 mRNA expression evolved in a stepwise fashion before the divergence of eutherian and marsupial mammals. A short G-rich element (G30) located downstream of the alternative 3'splice site of TRα2 mRNA and antisense to the 3'UTR of Rev-erbα plays an important role in regulating TRα2 splicing. G30 is tightly conserved in eutherian mammals, but is absent in marsupials and monotremes. Systematic deletions and substitutions within G30 have dramatically different effects on TRα2 splicing, leading to either its inhibition or its enhancement. Mutations that disrupt one or more clusters of G residues enhance splicing two- to three-fold. These results suggest the G30 sequence can adopt a highly structured conformation, possibly a G-quadruplex, and that it is part of a complex splicing regulatory element which exerts both positive and negative effects on TRα2 expression. Since mutations that strongly enhance splicing in vivo have no effect on splicing in vitro, it is likely that the regulatory role of G30 is mediated through linkage of transcription and splicing.

Fig 1. Structure of the TRα/Rev-erbα locus.

(A) Schematic representation of the conserved exon structure of TRα1, TRα2 and Rev-erbα mRNAs in eutherian mammals. Exons and introns are indicated by thick boxes and horizontal lines, respectively, with exon numbers indicated above. Horizontal arrows indicate direction of transcription; AAA represents major poly(A) sites. TRα1 mRNA includes exons 1–9. TRα2 mRNA includes exons 1–8, 9A and 10. Constitutive splicing is indicated by the angled solid lines, and the angled dotted line represents the alternative splicing of TRα2 mRNA. Thin boxes labeled C_n, G_n, A/T, Y_n and G30 represent conserved regions enriched in the indicated nucleotides as described in text; the vertical red arrow indicates the minor poly(A) site for Rev-erbα; and the double-headed arrow indicates the region shown in Panel B. The stop codon for each mRNA is indicated with a vertical red line. (B) Alignment of sequences from four mammals extending from the 3’ end of TRα1 mRNA (on the strand shown here) to the 3’ portion of exon 8 of Rev-erbα mRNA on the complementary strand. Yellow shading indicates identical residues in all aligned sequences; blue shading indicates identical residues in at least half of the aligned sequences. Red boxes delineate conserved sequence elements: the hexanucleotide polyadenylation signals (PAS), the non-canonical AT-rich sequence upstream of the major Rev-erbα poly(A) site (A/T), the consensus PUF protein-binding site (PUM) in Rev-erbα mRNA, and an AU-rich element in Rev-erbα (ARE). TRα1 and Rev-erbα poly(A) sites in rat and human are marked with large horizontal red arrows, the TRα2 3’ss with a downward vertical red arrow and poly(A) sites in platypus Rev-erbα mRNA with upward vertical red arrows.

Fig 2. Evolution of the alternative 5’ss in exon 9 of TRα1/TRα2 mRNA.

(A) Splicing of chimeric pre-mRNAs in vitro. Structure of pre-mRNA constructs, indicated on diagram at top, includes exon 9 sequences (shaded box) from opossum, rat, platypus or chicken together with a portion of intron 9 and exon 10 from rat (open box). Rat(3G) is a construct in which three G residues are substituted in the rat pre-mRNA. Labeled pre-mRNAs were incubated with nuclear extract for the times indicated and analyzed by electrophoresis as shown on the autoradiogram. Arrows indicate unspliced pre-mRNA (top) and spliced mRNA, as shown by cartoons at left. Spliced mRNAs were confirmed by RT-PCR sequencing. Note: both rat pre-mRNAs include an additional 13 nt of vector sequence at their 5’ end that results in slower migration relative to the chimeric pre-mRNAs. (B) Alignment of sequences adjacent to the 5’ splice site of TRα2 in exon 9. Shading is similar to that in Fig 1B. Normal and cryptic 5’ splice sites are indicated with solid arrows at top with positions relative to the 3’ss for exon 9 given in parentheses. Consensus splice site sequences are boxed. Dotted arrows at bottom indicate residues within the normal or cryptic 5’ss that are conserved in all eutherian and marsupial species. Each of these residues, identified by their position relative to the 5’ss, is replaced with G in the Rat(3G) construct. Human and lizard sequences are shown for comparative purposes.

Fig 3. Analysis of conserved sequences required for alternative polyadenylation of Rev-erbα mRNA.

(A) Schematic showing structure of two Rev-erbα mRNAs alternatively polyadenylated at the minor (top) and major (bottom) sites as indicated in Fig 1A. Red box between the two poly(A) sites represents the position of the PUF protein consensus binding site (PUM) and vertical line the position of the cryptic poly(A) site. Large arrow at right indicates direction of transcription; small arrows indicate primers, with 3’ RACE primers indicated as bent pair of arrows. (B) 3’ RACE mapping of alternative poly(A) sites with the rat Rev-erbα minigene. PCR products from 3’RACE of mRNA from cells transfected with Rev-erbα minigenes were analyzed by gel electrophoresis. Lanes 1–4 show effect of mutations on the poly(A) signal sequences upstream of the major and minor poly(A) sites. Lanes 5–8 show effects of additional mutations. Arrows indicate three 3’RACE products corresponding to the major and minor alternatively polyadenylated Rev-erbα mRNAs and RNA polyadenylated at a cryptic site. (C) Sequence of 3’ end of Rev-erbα encompassing the poly(A) sites. Conserved elements targeted by mutations are underlined with the substituted nt indicated by small letters below; polyadenylation sites are marked with forward slashes. A non-conserved, AT-rich sequence upstream of the cryptic site is noted with dashed underlining.

Fig 4. Substitutions and deletions within the G30 region affect TRα2 splicing.

(A) Schematic representation of exon 10 of TRα2 showing substitutions within the G30 region. Diagram at top shows intron/exon structure of erbAm minigene (exons 7–10). Dark shading within exon 10 indicates bidirectional coding sequence (BCS) where coding sequences for TRα2 and Rev-erbα overlap; medium shading represents other coding sequence; light shading 3’UTR. Small red and blue arrows indicate positions of RT-PCR primers for TRα1 and TRα2 mRNAs, respectively. Lower diagrams indicate structure of exon 10 for wildtype (WT) and five mutations. Substitution of homologous sequences from either Rev-erbβ gene (βGF) or opossum Rev-erbα gene (POS) are indicated as hatched boxes; open box indicates deletion of G30 (ΔG30). TRα2 splicing, measured by real-time RT-PCR, is given at right as % of wildtype splicing (standard deviation), N = 3; wildtype splicing = 28% (SD 8.2%). Asterisks indicate the significance of the change in expression of TRα2 in mutant compared to WT as determined by Student’s t test (* p< 0.05, ** p < 0.01). (B) Sequences of the G30 region for wildtype and mutant constructs shown in panel A. Positions +33 to +63, as determined from 5’ end of exon 10, correspond to G30. Flanking sequences are shown, including BsrGI site (TGTACA) used in construction and sequence antisense to Rev-erbα stop codon (TCA). Underlined WT sequence corresponds to a predicted exonic splicing enhancer. Substituted nt are shown in small letters.

Fig 5. Role of G-clusters in G30 probed by deletion scanning mutations.

(A) Diagram of the structure of the ErbAm minigene as in Fig 4A. The longer line under exon 9 represents the riboprobe used for RNase protection assays in panel D, the shorter line RNase protected probe annealed to spliced TRα2 mRNA. (B) Sequences of closely spaced 12 and 18 nucleotide deletions (ΔG12, ΔG18) are indicated within the G30 region. Deletions are named according to the position of their 5’ ends in exon 10. Clusters of three or more G residues are underlined. The possible 8-nt splicing enhancer sequence is highlighted in the wildtype sequence. The positions of some deletions are ambiguous, and these are shown twice to emphasize stepwise positioning of the deletions. TRα2 splicing determined by real-time RT-PCR is shown at right as % of wildtype splicing with standard deviation given in parentheses (N = 3–6); WT = 25% TRα2 splicing (SD 6.2%). (C) Bar graph summarizing results from ΔG12 deletions in panel A. Ambiguous positions are indicated with two identical bars; blue bars indicate deletions which disrupt at least one G3 cluster. One or two asterisks indicate the significance of the change in expression of TRα2 in mutant compared to WT (p < 0.05 or p <0.01, respectively). (D) Autoradiogram showing results from RNase protection assays carried out in parallel using a probe complementary to the TRα2 5’ss in exon 9. Minigene deletions are labeled similarly to panel B with ambiguous positions indicated with forward slash.

Fig 6. Substitutions within G30 support a critical role for G clusters in regulating TRα2 splicing.

Substitutions within the G30 region of minigene ErbAm are shown with results of real-time RT-PCR assays as in Fig 5B (N = 3–6).

Fig 7. Mutations in the G clusters of G30 have no effect on splicing of TRα2 pre-mRNA in vitro.

Four mutations that greatly enhance splicing of rat TRα2 minigenes when expressed in vivo were incorporated into the pα2- ΔBS minigene. Rat TRα2 pre-mRNAs with or without four mutations described in Figs 4–6 were incubated under splicing conditions as shown in Fig 2. Small arrows indicate unspliced RNAs at top and spliced products at bottom. Percent splicing is calculated from phosphoimager scans after correcting for nucleotide composition of the spliced and unspliced RNAs.