/Products/RNAi (RNA Interference)/shRNA/Design
Products > RNAi (RNA Interference) > shERWOOD-UltramiR shRNA > Design

shERWOOD-UltramiR shRNA

shERWOOD-UltramiR shRNA Design

The shRNA-specific shERWOOD algorithm designs combined with an optimized UltramiR microRNA scaffold provide increased and consistent knockdown efficiency (Knott, et al., 2014).  An unbiased screen (“Sensor Assay”) of 270,000 shRNA sequences was used to train the shERWOOD algorithm.  Of these, only ~2% of the sequences tested showed extremely potent knockdown at single copy and this data set was used to train the shERWOOD algorithm.  In addition, optimization of the microRNA scaffold provides increased microRNA processing.  The figures below outline the screen and provide examples of efficient knockdown and increased processing.  

Figure 1. Schematic showing the sensor assay used to validate 270,000 sequences and train the shERWOOD algorithm.

High-throughput sensor assay used to train the shERWOOD algorithm

The Sensor Screen tested shRNA knockdown efficiency using sequences inserted into a primary miR-30 scaffold so as to undergo microRNA pathway processing. shRNA expression was under the control of a doxycycline-inducible promoter in a viral vector that contained the target of the shRNA fused to a green fluorescent reporter gene (Venus).  Fluorescence could then be used as a “Sensor” to separate cells expressing with efficient shRNA from those with inefficient shRNA, Figure 1 (Knott et al 2014).  

Over 250,000 shRNA targeting all genes in the human genome were functionally tested in a Sensor screen.   By analyzing the dropout rate of shRNA-mir at each step of microRNA processing (primary, precursor and mature microRNA) Fellmann et al 2011 showed that each shRNA processing had specific sequence biases that impacted both the rate and accuracy of processing and therefore potency of the hairpin.
 Sequence analysis and thermodynamic information from the shRNA was used to train the shERWOOD shRNA design algorithm (Knott et al 2014). 
•       The first shRNA-specific design algorithm
•       Optimized to predict designs based on potent single copy knockdown
•       Designs target all transcripts of the gene
•       Includes filters to minimize off target effects

shERWOOD shRNA designs provide potent knockdown even at single copy

Figure 2. Western blot (A) and graph (B) showing protein knockdown produced by several shERWOOD predicted hairpins targeting 3 genes. Cells were transduced at single copy (MOI=0.3) in HEK293T (A) or U2OS (B) cells. Dotted line represents 70% knockdown.

shERWOOD designs provide knockdown even when expressed from a single integration in the target cell.  The figure above shows knockdown at the protein level in HEK293T or U2OS cells after single copy transductions targeting PTEN, FANCA or FANCI. Top ranked hairpins targeting each gene produced effective and consistent protein knockdown.

Figure 3.  Relative abundances of processed guide sequences for two shRNA as determined by small RNA cloning and NGS analysis when cloned into traditional miR-30 and UltramiR scaffolds. Values represent log-fold enrichment of shRNA guides with respect to sequences corresponding to the top 10 most highly expressed endogenous microRNA.

Optimized scaffold for increased small RNA processing

Previous generation microRNA-adapted shRNA libraries have alterations in conserved regions of the mir-30 scaffold that were suboptimal for small RNA processing and consistency of knockdown.   The alternate miR scaffold called UltramiR has been optimized based on recent knowledge of the key determinants for optimal primary microRNA processing (Auyeung et al. 2013).  This new scaffold increases shRNA processing presumably by improving biogenesis. When shRNA were placed into the UltramiR scaffold, mature small RNA levels were increased roughly two fold relative to levels observed using the standard miR-30 scaffold (Knott et al., 2014).

shERWOOD-UltramiR hairpin

shERWOOD shRNA are expressed with the optimized ultramiR scaffold.  The figure below shows the shRNA secondary structure and highlights the sequences that are included in the mature RNAi trigger bound to the targeted mRNA.

shERWOOD-UltramiR shRNA Design
Figure 1. Schematic of shERWOOD-UltramiR shRNA. (A) Passenger (green) and Guide (orange) strand are shown with Dicer and Drosha nuclease cleavage sites are in red. (B) The final step of shRNA processing loads the Guide Strand (orange) into the RISC complex which binds the target mRNA (blue) in a sequence specific manner.   

Figure 4. Example  target regions for a single transcript gene (top) and two multiple transcript genes (middle and bottom). For the middle gene, a target region (composed of >250 bp present in >80% of transcripts) was identified on the first algorithm iteration. For the bottom gene, a second algorithm iteration was required, where the smallest transcript was not considered.  C) Example shRNA off-target algorithm implementation.  In case A, all rank 1-4 are non-multimappers, so no shuffling occurs. In case B, all rank 1-8 shRNAs are multimappers (indicting that the gene is a paralogue), so no shuffling occurs. In case C, some but not all rank 1-4 and rank 5-8 shRNAs are multimappers and shuffling occurs to select a set of 4 shRNAs that include the highest scoring non-multimappers. (Adapted from Knott et al 2014)

Optimal design space: limit off-target effects while providing the most complete knockdown

Many genes have multiple transcript variants expressed from the same loci.  Even the most potent shRNA may allow significant gene expression if only a subset of spliceforms are targeted.  The shERWOOD algorithm prioritizes designs targeting sequences shared by all know transcripts for a given gene. In addition, designs that target additional gene sequences  were deemed multimappers and were penalized to select for the designs that provide the most complete knockdown while limiting off-targets.