Targeting a pre-mRNA structure with bipartite antisense molecules modulates tau alternative splicing

Abstract

Approximately 15% of human genetic diseases are estimated to involve dysregulation of alternative pre-mRNA splicing. Antisense molecules designed to alter these and other splicing events typically target continuous linear sequences of the message. Here, we show that a structural feature in a pre-mRNA can be targeted by bipartite antisense molecules designed to hybridize with the discontinuous elements that flank the structure and thereby alter splicing. We targeted a hairpin structure at the boundary between exon 10 and intron 10 of the pre-mRNA of tau. Mutations in this region that are associated with certain forms of frontotemporal dementia, destabilize the hairpin to cause increased inclusion of exon 10. Via electrophoretic mobility shift and RNase protection assays, we demonstrate that bipartite antisense molecules designed to simultaneously interact with the available sequences that immediately flank the tau pre-mRNA hairpin do indeed bind to this structured region. Moreover, these agents inhibit exon 10 splicing and reverse the effect of destabilizing disease-causing mutations, in both in vitro splicing assays and cell culture. This general bipartite antisense strategy could be employed to modulate other splicing events that are regulated by RNA secondary structure.

Figure 1.

Complementary bipartite ASOs can bind to the discontinuous sequences that flank the tau hairpin. (A) Diagram showing predicted RNA folding at the exon 10 (uppercase, blue): intron 10 (lowercase, red) boundary to form a hairpin secondary structure. The design of a bipartite ASO against the sequences flanking the hairpin is shown underneath in black. (B) EMSA to show the extent of hybridization of DNA ASOs to the tau hairpin-containing RNA sequence. ASOs were designed against the discontinuous sequences that flank the hairpin or against the unfolded linear sequence. A scrambled sequence was also tested. (C) EMSA showing the extent of hybridization of DNA ASOs designed separately against the 5′- and 3′-regions that flank the tau hairpin RNA. Note that unhybridized 5′-flank ASO is too small to be detected. (D) EMSA revealing the importance of binding of the 3′-flank. The hybridization of DNA ASOs with 7 or 8 bases complementary to the 5′-flank, and 0–8 bases complementary to the 3′-flank was examined.

Figure 2.

RNA and PNA complementary ASOs can bind to the discontinuous sequences that flank the tau hairpin. (A) The extent of hybridization of RNA ASOs to the tau hairpin-containing RNA was tested by EMSA. Modified RNA does not stain with SYBR Gold, so bands representing unhybridized RNA ASOs are not seen on the gel. (B) EMSA showing the time-course of hybridization of 9-2A-10 and E10B RNA ASOs to the tau hairpin RNA. (C) Quantification of the intensity of the tau RNA band. Data are the mean ±  S.E.M. from three independent experiments, normalized to the intensity of tau hairpin RNA alone. (D) EMSA showing the effect of increasing flank length of PNA ASOs on the extent of hybridization with the tau hairpin-containing RNA. PNA ASOs also contain lysine residues to aid cell penetration for experiments beyond the scope of the current study.

Figure 3.

RNase digestion profiles of the 5′-biotinylated tau hairpin-containing RNA with bound DNA ASO. (A) RNase A 5′-digestion fragments of tau hairpin-containing RNA with and without protection by bipartite 9-2A-10 DNA. Digestion fragments were run alongside an alkaline hydrolysis ladder of all possible 5′-biotinylated fragments and alongside undigested RNA. Bands representing cleavage in the 5′-flank (below C−3) are too small to be resolved or detected on the gel. (B) RNase T1 digestion profile of the tau RNA as in panel A. (C) RNase H digestion of the tau RNA when bound to DNA ASOs. Digestion products were analyzed by denaturing urea/acrylamide gel electrophoresis and visualized by SYBR Gold staining under UV light.

Figure 4.

In vitro splicing of a tau splicing unit comprising exon 10, a shortened intron 10 and exon 11 in the absence or presence of a bipartite RNA ASO. Products from the in vitro splicing reactions were purified by phenol/chloroform precipitation and amplified by reverse transcription PCR. Unspliced RNA and spliced products were visualized by 8% acrylamide gel electrophoresis and SYBR Gold staining under UV light. In vitro splicing of a wild-type tau splicing unit (A) and a tau splicing unit containing the FTDP-17-associated DDPAC mutation (B) was inhibited by 9-1A-10 RNA in a concentration-dependent manner.

Figure 5.

The effect of RNA ASOs on alternative exon splicing from a tau minigene and from a BACE1 minigene expressed in SK-N-SH cells. (A) Diagram of alternative splicing pathways of a minigene encoding tau exons 9, 10 and 11. (B) Exon-trapping PCR from the wild-type minigene co-expressed with 9-1A-10, E10B, scrambled RNA antisense or no antisense in SK-N-SH cells. PCR products were analyzed by 8% acrylamide gel electrophoresis and SYBR Gold staining under UV light. (C) Exon-trapping PCR from the DDPAC minigene expressed in SK-N-SH cells. (D) RNA ASOs against the 5′- or 3′-flank of the tau hairpin (alone or in combination) have no effect on splicing from the wild-type or DDPAC (E) minigene. (F) Diagram of alternative splicing pathways of a minigene encoding BACE1 exons 3 and 4. (G) Exon-trapping PCR from the BACE1 minigene.

Figure 6.

Bipartite RNA ASO inhibits exon 10 splicing from a tau minigene expressed in SK-N-SH cells and from the endogenous tau pre-mRNA in HEK-293 cells, as quantified by real-time PCR. 10 nM 9-1A-10 RNA or E10B RNA caused a significant reduction in exon 10 splicing from the wild-type minigene (A), and from the DDPAC minigene (B), with a concomitant increase in the 3R isoform compared with 10 nM scrambled control RNA. (C) The inhibition of exon 10 splicing by 9-1A-10 RNA was concentration-dependent, with an IC₅₀ of 5.4 nM. (D) 9-1A-10 RNA ASO does not alter total tau expression from the tau minigene in SK-N-SH cells. Total tau expression from the minigene in transfected cells was normalized to the expression of pRL-TK, which was used as a transfection control. (E) Transfection of HEK-293 cells with 1 µM 9-1A-10 or E10B antisense RNA significantly inhibits exon 10 splicing from the endogenous tau pre-mRNA in comparison with transfection with a scrambled control. (F) Total endogenous tau is unaltered in cells transfected with 1 µM 9-1A-10, E10B or scrambled RNAs. Expression of total tau was normalized to expression of GAPDH.