Demonstrating polymorphic miRNA-mediated gene regulation in vivo: application to the g+6223G->A mutation of Texel sheep

Abstract

We herein describe the development of a biochemical method to evaluate the effect of single nucleotide polymorphisms (SNPs) in target genes on their regulation by microRNAs in vivo. The method is based on the detection of allelic imbalance in RNAs coimmunoprecipitated with AGO proteins from tissues of heterozygous individuals. We characterize the performances of our approach using a model system in a cell culture, and then apply it successfully to prove that the 3'UTR g+6223G-->A mutation operates by promoting RISC-dependent down-regulation of myostatin (MSTN) in skeletal muscle of Texel sheep.

FIGURE 1.

(A) Texel sheep with muscular hypertrophy resulting in part from the g+6223G-A mutation in the 3′UTR of the MSTN gene. The approximate positions of the semimembranosus (SM) and longissimus dorsi (LD) muscle examined in this study are shown. (B) Schematic representation of the allelic imbalance test developed in this study to demonstrate the preferential association of the mutant MSTN A allele over the wild-type MSTN G with the RISC complex. Prior to immunoprecipitation, transcripts carrying the G allele are predicted to be slightly more abundant than those carrying the A allele in skeletal muscle of heterozygous A/G animals as a result of the RISC-dependent degradation of the targeted A transcripts. The coimmunoprecipitation of miRNA-regulated mRNA with antibodies directed against RISC components is predicted to cause a net enrichment in targeted A transcripts.

FIGURE 2.

(A) Schematic representation of the luciferase and miRNA-expressing vectors used to optimize the allelic imbalance test in HeLa cells. pRL-TK-4xA and pRL-TK-4xG correspond to vectors expressing the renilla luciferase gene under the dependence of the HSV-TK promoter, endowed in their 3′UTR with four tandem copies of an 80-bp fragment encompassing, respectively, the mutant A and wild-type G residue at the g+6223G-A site. The pGL3 vector expresses firefly luciferase used to normalize for transfection efficacy. The pcDNA3-miR-x vectors drive expression of the indicated ovine pri-miRNAs (miR-1, miR-1M, and miR-377) from the CMV promoter. We performed the allelic imbalance test by transfecting equimolar amounts of pRL-TK-4xA and pRL-TK-4xG with either one of the pcDNA3-miR-x vectors and measuring the ratio of A to G allele in reporter-derived transcripts by direct sequencing of RT-PCR products before and after RIP. (B) Sequence complementarity between miR-1/miR-1M and the mutant/wild-type ovine MSTN alleles, respectively. miR-1M was derived from miR-1 by mutagenizing the U at position six into a C to make its seed perfectly complementary to the wild-type MSTN allele. While the U at position six of miR-1 forms a wobble pair with the matching G residue of the wild-type MSTN 3′UTR, the corresponding C residue of miR-1M and A residue of the mutant MSTN 3′UTR are mismatched. (C) Demonstration of the preferential down-regulation of pRL-TK-4xA by miR-1 (approximately threefold) and of pRL-TK-4xG by mir-1M (approximately fourfold). Renilla-to-firefly luminescence ratios observed when cotransfecting HeLa cells with reporter vectors as indicated and vectors expressing either miR-1 or miR-1M. Error bars correspond to 1.96 x SEM of two biological replicates. (D) Reduction of luciferase activity after cotransfection of HeLa cells with (1) an equimolar mixture of pRL-TK-4xA and pRL-TK-4xG, and (2) vectors expressing either miR-1, miR-1M, or the control miR-377. Ratios of renilla-to-firefly luminescence are reduced by approximately half (which is slightly more than expected and could reflect low-level cross-talk between pRL-TK-4xA and miR-1M and pRL-TK-4xG and miR-1, respectively), with a slightly stronger effect of miR-1M, confirming findings shown in C. Error bars correspond to 1.96 × SEM of three biological replicates. (E) Immunoprecipitation of endogenous human AGO proteins using the 2A8 anti-hAGO2 monoclonal antibody. HeLa cells were cotransfected with equimolar amounts of the two luciferase reporter vectors and one of the miRNA expression vectors (expressing either miR-1, miR-1M, or miR-377) as before. Cell lysates (without fixation) were incubated with either the 2A8 antibody or a nonimmune mouse IgG as a negative control. Two incubation times (either 2 h [h] or overnight [o.n.]) were tested. Input, supernatant (Sup) after immunoprecipitation, and immunoprecipitates (IP) were loaded on PAGE gels and subjected to Western blotting using the same 2A8 antibody, allowing detection of hAGO proteins (arrow, ∼95 kDa). Enrichment of hAGOs in IP samples and depletion in supernatants were observed when using the 2A8 antibody, but not control IgG. Cross-reaction with radixin (*) (Nelson et al. 2007) and elution of IgG heavy (h) and light (l) chains were observed. (F) Allelic imbalance test showing the preferential coimmunoprecipitation of the miRNA targeted alleles. RNA was isolated from the input and IP samples, reverse transcribed using random hexamers, and PCR amplified using primers spanning the tandem insertions. SYBRsafe-stained RT-PCR products after agarose gel electrophoresis are shown at top. The PCR products were directly sequenced. The electropherograms at the polymorphic site are shown in the second panel. Peak heights of each allele (A,G) were obtained using Peakpicker (Ge et al. 2005). The degree of allelic imbalance was estimated as log₂ of the ratio of ratios of A over G peak height, after (IP) and before immunoprecipitation (Input, third panel). Exact log₂ values are given for each vector combination. The miR-1/miR-1M and 2A8-dependent reciprocal allelic imbalance is clearly visible. (G) Schematic representation of the pRL-TK-3′A and pRL-TK-3′G reporter vectors endowed with the full-length, respectively, mutant and wild-type ovine MSTN 3′UTR sequences. (H) Testing the effect of number of miRNA target sites (four versus one) on the preferential IP of the targeted versus nontargeted allele. HeLa cells were transfected with equimolar mixes of pRL-TK-4xA/pRL-TK-4xG or pRL-TK-3′A/pRL-TK-4′G, and either pcDNA3-miR-1, pcDNA3-miR-1M, or pcDNA3-miR-377. Cells were fixed and lysed by sonication. RNA was isolated from the immunoprecipitate, reverse transcribed using random hexamers, and PCR amplified using primers spanning the tandem insertions (pRL-TK-4xA/pRL-TK-4xG) or g+6223G-A site (pRL-TK-3′A/pRL-TK-4′G). SYBRsafe-stained RT-PCR products after agarose gel eletrophoresis are shown. The PCR products were directly sequenced. The electropherograms at the polymorphic site are shown. The degree of allelic imbalance was estimated as log₂ of the ratio of ratios of A over G peak height with miR-1/miR-1M and with miR-377 (control). The miR-1/-1M-dependent reciprocal allelic imbalance is clearly visible at comparable levels for the vectors endowed with four versus one target-site copy.

FIGURE 3.

(A) Schematic representation of the experiment conducted to evaluate the rate of post-lysis reassociation of the RISC, which might generate artifactual evidence of polymorphic miRNA-mediated gene regulation. Plasmids expressing targets (mix of pRL-TK-4xA and pRL-TK-4xG) and miRNAs (either pcDNA3-miR-1 or pcDNA3-miR-1M) were either transfected together (T) or separately (S) in HeLa cells. The plasmid mixes were complemented with matching empty vectors (pRL-TK, pcDNA3) to ensure transfection of equal amounts of DNA, as indicated. Cells were either fixed with formaldehyde (0.5%, room temperature, 10 min) or not, and complementary cell populations mixed as indicated before proceeding with the AGO-RIP procedure. (B) Allelic imbalance test demonstrating that post-lysis reassociation of the RISC is negligible in the experimental conditions used. RNA was isolated from the input and IP samples, reverse transcribed (RT+) or not (RT−) using random hexamers, and PCR amplified using primers spanning the tandem insertions. SYBRsafe-stained PCR products after agarose gel eletrophoresis show the RT-PCR products corresponding to the pRL-TK-4xA/pRL-TK-4xG as well as empty pRL-TK vectors, respectively. The pRL-TK-4xA/pRL-TK-4xG-derived RT-PCR products were directly sequenced. The electropherograms at the polymorphic site are shown for both input and IP. Peak heights of each allele (A, G) were obtained using Peakpicker. The degree of allelic imbalance was estimated as log₂ of the ratio of ratios of A over G peak height, after (IP) and before immunoprecipitation (Input). The miR-1/-1M and 2A8-dependent reciprocal allelic imbalance is observed only when target- and miRNA-expressing vectors are transfected together, not when transfected separately, irrespective of fixation conditions.

FIGURE 4.

(A) Immunoprecipitation of ovine AGO proteins using the 2A8 anti-hAGO2 antibody. Skeletal muscle samples (longissimus dorsi and semimembranosus) from a 5- and a 30-wk-old sheep were minced, fixed in 1% formaldehyde or not, deep-frozen, and lysed. Cell lysates were incubated overnight with the 2A8 antibody. (Input), supernatant (Sup) after immunoprecipitation, and immunoprecipitates (IP) were loaded on PAGE gels and subjected to Western blotting using the same 2A8 antibody. Enrichment of ovine AGOs in IP samples and depletions in supernatants were observed. (B) Allelic imbalance test performed on the fixed and nonfixed skeletal muscle lysates of the 30-wk-old animal, heterozygous for the g+6223G-A mutation (the 5-wk-old animal was homozygous wild type). RNA was isolated from the input and IP samples, reverse transcribed (RT+) or not (RT−) using random hexamers, and PCR-amplified using primers spanning the g+6223G-A site. SYBRsafe-stained PCR products after agarose gel electrophoresis are shown, and were directly sequenced. The electropherograms at the polymorphic site are shown for both input and IP. Peak heights of each allele (A, G) were obtained using Peakpicker. The degree of allelic imbalance was estimated as log₂ of the ratio of A over G peak height. Note the depletion of the mutant A allele in the input as a result of its miR-1/-206-dependent degradation, yet its clear enrichment after AGO-RIP, demonstrating its preferential association with the RISC in vivo. (C) Allelic imbalance test using fluorescent PCR-RFLP as alternative. Genomic and RT–PCR products obtained from a heterozygous sheep were labeled by adding a fluorescent primer at the last PCR cycle (“hot stop PCR”) (Uejima et al. 2000) and digested with a restriction enzyme cleaving the wild-type but not mutant allele. After capillary gel electrophoresis on an ABI3730 instrument, the mutant and wild-type allele were recognized as fluorescent fragments of, respectively, 253 and 98 bp. The areas under the allele-specific peaks were estimated using the Peak Scanner software (Applied Biosystems). The degree of allelic imbalance was estimated as log₂ of the ratio of A over G peak area in the RT-PCR product.