Design of RNAi hairpins for mutation-specific silencing of ataxin-7 and correction of a SCA7 phenotype

Abstract

Spinocerebellar ataxia type 7 is a polyglutamine disorder caused by an expanded CAG repeat mutation that results in neurodegeneration. Since no treatment exists for this chronic disease, novel therapies such post-transcriptional RNA interference-based gene silencing are under investigation, in particular those that might enable constitutive and tissue-specific silencing, such as expressed hairpins. Given that this method of silencing can be abolished by the presence of nucleotide mismatches against the target RNA, we sought to identify expressed RNA hairpins selective for silencing the mutant ataxin-7 transcript using a linked SNP. By targeting both short and full-length tagged ataxin-7 sequences, we show that mutation-specific selectivity can be obtained with single nucleotide mismatches to the wild-type RNA target incorporated 3' to the centre of the active strand of short hairpin RNAs. The activity of the most effective short hairpin RNA incorporating the nucleotide mismatch at position 16 was further studied in a heterozygous ataxin-7 disease model, demonstrating significantly reduced levels of toxic mutant ataxin-7 protein with decreased mutant protein aggregation and retention of normal wild-type protein in a non-aggregated diffuse cellular distribution. Allele-specific mutant ataxin7 silencing was also obtained with the use of primary microRNA mimics, the most highly effective construct also harbouring the single nucleotide mismatch at position 16, corroborating our earlier findings. Our data provide understanding of RNA interference guide strand anatomy optimised for the allele-specific silencing of a polyglutamine mutation linked SNP and give a basis for the use of allele-specific RNA interference as a viable therapeutic approach for spinocerebellar ataxia 7.

Figure 1. RNAi-based shRNAs and target vectors designed for atxn7 knockdown.

(A) Convention for the diagram follows Schwarz and colleagues (16). Sequences of expected single and double mismatch guide strands (underlined) processed from an shRNA format indicating the site of the primary mismatch to the G allele of the wild-type atxn7 target. Sequence context of both the wild-type and mutant targets are shown. Functional asymmetry was induced by creating G:U mis-pairings in the 3′ end of the anti-guide strand; these are highlighted in bold. The shRNAs were labelled according to the position of the primary mismatch relative to the 5′ end of the guide strand; followed by any additional mismatches. The nomenclature is based on a 21bp guide strand after cleavage of the anti-sense strand by Dicer. (B) Schematic representations of the fused target-reporter cassettes. atx7-G-luc and atx7-A-luc are the wild-type and mutant reporter siCHECK plasmids used in the luciferase assay, and included a short region of 60 bp of atxn7 target sequence spanning the SNP fused to the Renilla luciferase reporter gene. atx7-10-G-eGFP, atx7-100-A-eGFP, atx7-100-A-DsRED are the 3 reporter plasmids used in the full-length hemi- and heterozygous assays and consist of the full-length atxn7 cDNA. The wild-type construct (atx7-10-G-eGFP) has a CAG repeat length of 10 incorporating the G allele of the SNP, fused to eGFP. The two mutant constructs (atx7-100-A-eGFP, atx7-100-A-DsRED) have CAG repeat lengths of 100, as well as the A allele of the SNP, and are fused to eGFP and DsRED respectively.

Figure 2. Analysis of series of single mismatched shRNA guide sequences targeting the G>A SNP in atxn7 in a dual luciferase assay.

Relative levels of Renilla luciferase (hRluc) expression normalized to firefly luciferase (hFluc) expression for single mismatches (guide sequences shown in Figure 1A). Each experiment was performed in triplicate and the data is relative to that measured using a non-specific shRNA, shR-NS. Average±standard deviation is shown. Statistically significant differences (p<0.05) between wild-type and mutant silencing are indicated by corresponding p values. Relative expression of wild-type and mutant targets is represented by red and blue bars respectively.

Figure 3. Single mismatched shRNAs targeting the atxn7 G>A SNP in a full-length hemizygous assay.

Quantitation of fluorescence in HEK293 cells transfected with wild-type (atx7-10-G-eGFP) or mutant (atx7-100-A-eGFP) expression plasmids and the indicated shRNA is shown. Each experiment was performed in triplicate and the data is relative to that measured using a non-specific shRNA, NS. The shR-E19 hairpin was pE19; a published U6/shRNA expression plasmid targeting eGFP (37). Average±standard deviation is shown. Statistically significant differences (p<0.05) between wild-type and mutant silencing are indicated by corresponding p values. Wild-type and mutant targets are represented by red and blue bars respectively.

Figure 4. Knockdown of targets in a heterozygous assay using shRNAs.

Quantification of fluorescence in HEK293 cells co-transfected with wild-type atx7-10-G-eGFP and mutant atx7-100-A-DsRED expression plasmids and the indicated shRNA is shown. Each experiment was performed in triplicate and the data is relative to that measured using a non-specific shRNA, NS. Average±standard deviation is shown. The shR-E19 hairpin was pE19; a published U6/shRNA expression plasmid targeting eGFP (37). Statistically significant differences (p<0.05) between wild-type and mutant silencing are indicated by using the one-tailed t-test. Wild-type and mutant targets are represented by green and red bars respectively.

Figure 5. Investigation of aggregate formation.

(A) Representative confocal images of cells transfected with (i, ii) mutant (atx7-100-A-DsRED) alone; (iii, iv) wild-type (atx7-10-G-eGFP) alone. (i) Image visualized under red fluorescence, (ii) red fluorescence merged with the bright field, (iii) green fluorescence, and (iv) green fluorescence merged with the bright field. (B) Representative confocal images of cells co-transfected with wild-type and mutant expression plasmids in addition to (i–iii) shR-NS or (iv–vi) shR-P16. (i) and (iv) show images under green fluorescence to reveal wild-type protein, (ii) and (v) show images under red fluorescence to reveal mutant protein, and (iii) and (vi) show images then merged under green and red fluorescence and bright field. Scale bar represents 10 µm. C. Cells expressing eGFP and/or DsRED were counted according to whether they contained aggregates or a dispersed pattern of expression of mutant and wild-type ataxin-7. Cells were transfected with wild-type target (atx-10-G-eGFP) alone; mutant target (atx7-100-A-DsRED) alone; mutant (atx7-100-A-DsRED), wild-type (atx-10-G-eGFP) and shR-NS (non-specific shRNA); mutant, wild-type and shR-P16. Cells were counted separately in the red and the green filter by collecting 3 representative images from each well and combining the total number. This was performed for each indicated combination in biological triplicate, yielding standard deviations. The bars comprise the total number of cells counted in each transfection; separated according to whether they contained aggregates (blue) or a dispersed pattern of expression (grey). Note that the decrease in expression from target vectors transfected alone to co-transfected cells is due to a promoter occlusion effect of co-expression of targets and not due to the addition of shR-NS which has no effect upon the target vectors (data not shown). Statistically significant differences in % of aggregate containing cells are indicated by corresponding p values.

Figure 6. Knockdown of targets using pri-miRNA-based hairpins.

A) Guide strands in pri-miRNA format, with sequences of mismatch guide strands processed from a miRNA format indicating the site of the primary mismatch to the G allele of the wild-type atxn7 target. Sequence context of both the wild-type and mutant targets are shown. The miRNAs were labelled according to the position of the primary mismatch relative to the 5′ end of the guide strand. Major species refers to the strand most likely to be incorporated by RISC, while the minor species indicates the strand less likely to be incorporated. B) Knockdown of targets in a heterozygous assay using miRNAs. Quantification of fluorescence in HEK293 cells co-transfected with wild-type atx7-10-G-eGFP and mutant atx7-100-A-DsRED expression plasmids and the indicated miRNA. The data is relative to that measured using a non-specific miRNA, NS. Average±standard deviation is shown. Statistically significant differences (p<0.05) between wild-type and mutant silencing are indicated by using the one-tailed t-test. Wild-type and mutant targets are represented by green and red bars respectively. C. Increased addition of an shRNA but not a miRNA hairpin interferes with post-transcriptional gene silencing machinery. Low levels of a GFP target plasmid and GFP specific hairpin result in knockdown of GFP relative to a non-specific hairpin in the presence of 1 µg of empty vector, either pU6+1 or pCI_neo (vectors which contain the promoters used to transcribe shRNA and miRNA hairpins respectively). The shRNA data is relative to that measured using a non-specific shRNA, NS, while the miRNA data is relative to that measured using a non-specific miRNA, NS. The shR-E19 hairpin was pE19; a published U6/shRNA expression plasmid targeting GFP (37). Average±standard deviation is shown. Each experiment was performed in triplicate.