shRNA stable lines - no stable reduction seen (Jul/22/2009 )

I recently created a series of stable lines using an shRNA vector that also co-expresses GFP and an antibiotic resistance gene, but after stable line creation found no clones exhibiting knockdown of my gene of interest.

Here's how it went:
To begin, I verified by qRT-PCR and Western blotting following transient transfection that my shRNA vector was able knockdown expression of my gene of interest. Knockdown was at least 50% with 60-70% transfection efficiency after 72 h, with no evidence of cell death.
I then performed a kill curve on my target cells to determine the lowest concentration of antibiotic that would kill 100% of my cells within 1-4 days.
To generate the stable lines, I used a plasmid based method where I transfected my target cells as above with the shRNA vector, or a non-effective shRNA control, or mock transfected the cells. After 72 h of recovery time, I trypsinized my cells, and replated them in my previously determined concentration of antibiotic. The mock cells died within 1 day and never formed any colonies. The shRNA transfected cells formed nice looking distinct colonies within 1 week or so and I checked their fluorescence (the vector also contains GFP expression) as a means of loosely assessing clonality and expression level.
I picked at least 12 independent clones of each (effective and non-effective shRNA) and separately expanded each onto triplicate plates and harvested one of each for RNA and protein. While all the colonies showed clonal and high expression of GFP, no clones exhibited any knockdown of my gene of interest.

Again, I believe the construct is fine since it worked transiently in several experiments.
Does anyone have any insight into what could be going on here?

I was thinking that perhaps the copy number of the plasmid is low in the stable vs the transient transfection, or maybe reduction of my gene of interest is detrimental to the cells leading to selection against those with reduced levels of expression.
Is there anything else that could be going on?

Dr Teeth on Jul 22 2009, 09:38 AM said:

I think you have already thought about the most common explanations: stable copy number and counter-selection of knockdown cells.

I have also faced this situation in the past where constructs giving 100% knockdown in transient assays, gave about 20-30% knockdown after generating stable cell lines, and I had to eventually give up on it.

Some things you can do, however, if you must make these stable cell lines and have a time on hand...

(1) Try out higher conc of antibiotics (not minimum necessary to kill 100% cells). Logic being that if some clones do survive such high conc, they may have higher integration numbers and probably more knockdown efficiency.

(2) Design new siRNA constructs, both using different region of mRNA, as well as different siRNA vectors.

(3) If cell death due to gene knockdown looks like the real trouble, you can overcome this by making inducible stable knockdown cell lines (such as pSingle).

HTH/

If I understand it correctly, I think your problem is "not linearise your plasmid before transfection". The non-homologous end joining increases the chance of integrating your drug selection marker along with the shRNA cassete. Without linearising plasmid DNA, maybe less than 40% of your selected cells have the shRNA casste integrated. The math is 0.4 (shRNA positive cells) X 0.5 (knockdown efficiency) = 0.2, your overall knock down efficiency is less than 20%.
Also, I will suggest you to pool cells instead of working on single clones. I have seen people worked years and finally realized the phenotype/results came from the clonal effect.
Hopefully it helps.

Functional Screens on Jul 22 2009, 06:53 PM said:

"Functional screens": I tend to disagree on two points. While not linearizing the DNA will reduce integration, since I am selecting for cells that have integrated the shRNA cassete, I don't think it matters. As I said, my cells survive puromycin treatment and express GFP which is part of the shRNA vector, so it is unlikely I think that the shRNA cassete would not have also been integrated. While the breakpoints might occur in a fashion that disrupts production of the shRNA, I evaluated over 20 clones from each and it is unlikely that this would have happened in each clone. Secondly, I disagree on the pooling. Pooling your cells leads to dealing with cells with a wide variety of different mutations/integration sites and since they are non-clonal, the cells will likely change over time as some sub-populations may outcompete others, leading to laboratory selection that would not be replicated in other labs. While single clones surely may have clonal effects that are specific to them as you pointed out, that is why it is always necessary to analyze multiple clones for each line to determine if your observed effects are merely clone dependent or are the result of your manipulation.


At any rate, my problem is now moot as it appears that the lack of reduction of my target was due to cell lethality resulting from loss of my protein of interest over time. A second group just published a paper showing that stable lines generated using shRNA against my gene all die after 1 week, while transient reduction using siRNA or inducible shRNA does not lead to cell death, unless maintained over time also.

Thanks for the help, though, everyone!