Guide on how to design shRNAs according to the paper “Short hairpin RNAs (shRNAs) induce sequence-specific silencing in mammalian cells” (Paddison et al. ,2002)

Be aware, that with shRNAs you never achieve that the gene is completely off – you typically achieve a reduction of gene expression by 20-80% per shRNA. Also, standard value: you probably have to design a few shRNAs so that one might work. CRISPR just works, that’s why people love it. But CRISPR deletes the gene, so if complete knockout is lethal for the cell (organism) then shRNAs might be your better fit.

In „lower“ organsims such as nematodes and plants, genes can be silenced by RNAs that are complementary to the mRNA of the gene. Scientists found that out by just expressing long antisense (messenger) RNAs – for example if the native gene is

ATGATTAGAAAAAAAAAAA then you express the antisense RNA

TACTAATCTTTTTT (“the reverse complement”) and this binds to the RNA and prevents the ribosomes from translating proteins from the RNA template.


However, human cells commit apoptosis when they encounter long double-stranded RNA (antiviral protection for the organism). The silencing RNA binds to mRNA and thus blocks translation, but also other mechanisms have evolved where proteins such as Drosha produce the small silencing RNAs from a template (naturally, from a viral dsRNA). Without going any deeper into how the silencing works – highly recommended to read the paper and follow up literature – we want to do some practical work now.

So in the following, we want to design a piece of DNA that gets transcribed into shRNAs and silences a gene of interest.

In this graphic from the paper we see that easy shRNA are the most effective. No sequence mismatches are needed.

Standard length is 29 nucleotides for the shRNAs, but can be reduced to 25 bp without big losses.

In the first 5 pages of the paper, they analyse the efficiency of chemically synthesized RNAs, then they try out to have the cells produce them in vivo.

Using the CMV promoter to drive hairpins didn’t work, as the shRNAs produced would have polyA tail and 5’ cap and would be exported out of the nucleus (where the enzymes work). Also significant extensions of any strand reduced shRNA activities.

Therefore, class III promoter such as U6 can be used, they don’t produce any extensions of the RNAs.


Basically this should be all the information you need to design your own shRNAs.

U6 Promoter cassette from pX330:

U6 Promoter
CRISPR binding part of the RNA

U6 promoter will start with the first G (in the spacer) and stop at TTTTTT. CRISPR part are not needed for shRNAs.



Here a picture of how such a thing could look

snapgene shRNAs

the cells which carry this DNA will make the RNA hairpin loop, and process it into silencing RNAs loaded onto the respective silencing proteins. That’s probably  all you need to know to work with it on an abstract level