An optimized microRNA backbone for effective single-copy RNAi

Compared with endogenous human MIR30A, the experimental miR-30 shRNAmir backbone differs in three conserved features: its synthetic stem has no bulge and carries the guide on the opposite strand, the two conserved base pairs flanking the loop are UA/UA instead of CU/GG, and the XhoI/EcoRI cloning sites mutate a highly conserved region 3′ of the basal stem.

The miR-E backbone restores the conserved region 3′ of the basal stem by moving the EcoRI cloning site to a nonconserved position further 3′ and altering the 5′ miR-30 scaffold opposite it in a way that minimally changes secondary structure. In a single-copy dTomato reporter assay in NIH 3T3 cells, miR-E raises knockdown by the intermediate shRNA Pten.1524 (62% protein knockdown in miR-30) into the range of potent control shRNAs such as Pten.1523 (96% in miR-30).

The miR-R1B1 backbone, designed to resemble the endogenous MIR30A stem-loop, raises Pten.1524 knockdown to a similar degree as miR-E, but miR-E is preferred because existing miR-30 reagents convert to it by PCR subcloning. Reversed passenger-guide configurations such as miR-R1B1 leave the 5′ end of the guide to Drosha cleavage, which is less precise than Dicer cleavage, which cuts miR-30 at the intended site with 98% accuracy.

Converting Mcl1, Bcl2 and Pten shRNAs of different strengths from miR-30 to miR-E increases single-copy knockdown in almost all cases and turns intermediate shRNAs into potent RNAi triggers. For Pten shRNAs the gain is comparable in NIH 3T3 fibroblasts, BNL CL.2 hepatocytes, SVEC4-10 endothelial cells and TCMK-1 kidney cells, and similar improvements occur in chicken embryonic fibroblasts.

Seven validated shRNAs against genes required for MOLM-13 acute myeloid leukemia cells deplete shRNA-expressing cells faster in competitive proliferation assays when expressed from miR-E than from miR-30. The control shRNA Ren.713 has no effect in either backbone, so miR-E does not cause general toxicity.

Eleven Dnmt3a shRNAs designed with up-to-date Sensor rules give variable knockdown in miR-30, with only two as potent as the control shRNA Ren.713, but all 11 trigger more than 80% single-copy knockdown in miR-E. The authors report similar results for more than 50 further shRNAs against other genes and no case in which miR-E had a negative effect.

A conserved -ACNNC- motif 3′ of the basal stem is required for miR-E function. Point mutations at three conserved positions reduce miR-E Pten.1524 knockdown to or below miR-30 levels, mutations at three further positions reduce potency detectably only by western blot, and the two nonconserved positions tolerate mutation. Three further conserved nucleotides extend the motif to -ACNUCAA-.

Endogenous human microRNAs are enriched for (non-C)CNNC motifs (ACNNC, GCNNC, UCNNC) in the 3′ flank of the Drosha cleavage site, and miRNAs carrying these motifs have higher mature miRNA levels in HEK293T cells. This indicates that the motif restored in miR-E reflects a general feature of efficient miRNA biogenesis.

The miR-E backbone increases Drosha–DGCR8 processing of the primary transcript. At single copy, GFP–miR-E transcripts give lower GFP fluorescence and miR-E retroviruses package at lower titers than their miR-30 counterparts, and siRNA knockdown of DROSHA or DGCR8 raises GFP from both backbones to similar levels. In NIH 3T3 cells miR-E lowers pri-miRNA levels and raises mature small RNA 7- to 32-fold, about tenfold for the potent shRNAs Ren.713 and Pten.1523.

Single-copy miR-E shRNAs produce mature small RNA levels well below those of the most abundant endogenous miRNAs and leave endogenous miRNA abundance unchanged relative to miR-30 (Pearson r = 0.99 across endogenous miRNAs in NIH 3T3 cells).

For three Pten shRNAs in miR-30 and miR-E, Dicer cleaves the pre-miRNA at its canonical site in 82–100% of reads, whereas Drosha cleaves at its canonical site in only 25–70% of reads, with many cleavages 1–2 nucleotides closer to the loop. Placing the guide on the 3′ stem side therefore lets Dicer fix the 5′ end of the guide, which determines the seed, while Drosha imprecision only varies the guide 3′ end.

Adding the endogenous loop (L), R1B1 or R2B2 stem-loop features to miR-E does not further improve knockdown by the already potent miR-E. The same features individually enhance knockdown by miR-30 Pten.1524.

Existing miR-30 shRNAs convert to miR-E by PCR amplification with miR-E primers followed by XhoI/EcoRI cloning, and de novo shRNAs are made by PCR amplification of 97-mer oligonucleotides. Combined with Sensor-rule shRNA designs, more than 50% of the top predictions for a gene trigger more than 80% protein knockdown at single copy in miR-E.

Constitutive and Tet-regulated miR-E expression vectors have been built in retroviral (pMSCV, pQCXIX) and lentiviral (pRRL) backbones. The Tet-regulated versions express fluorescence-coupled miR-E shRNAs from the T3G promoter, which strongly reduces leaky shRNA expression, and the rtTA3 variants add a positive feedback loop for single-vector Tet-ON use.