I understand you correctly, you're using a single plasmid or multiple plamids expressing Cas9, sgRNA,  GFP and antibiotic resistance marker to mutate by NHEJ and so induce a frame-shift in your protein to knock it out.

If that is correct, then there are a few options, but all of them involve cloning your cell lines. Since you have the GFP/drug resistance marker you could first sort or select on the basis of that to get rid of any untransfected cells. You could then clone some cell lines by limiting dilution or picking colonies. Identifying clones with both alleles knocked out would be simplest if you had an antibody against your protein of interest since you could use immunofluorescence or Western blotting. That way you're directly measuring the result you want ie. loss of protein. If you haven't got an antibody, you could amplify the targeted locus by PCR and sequence it for some of your clones. You would have to interpret the mixed signal on the sequencing trace. This is actually easier than it sounds although there are some online tools which will attempt to do it automatically (see link below). One caution I would make, don't assume your cell line is diploid. Most immortalised cell lines have pretty weird karyotypes, so you may have to knockout 3 or more alleles.

If you want or need to avoid cloning your cell line, then it really depends on the efficiency of your transfection and sgRNA. If you get these high enough, you may even find that the majority of your cells are knocked out. Again, immunofluorescence would be a good way to judge that. If the efficiency is high enough then you can avoid cloning and the risks that come from studying a few clonal lines which may not be representative of your cell line. I don't have much experience of using CRISPR with NHEJ to know what the efficiency is like. I've only been using it for homologous recombination which has a much lower efficiency.

Which brings me onto another option with a different strategy. You could try inserting you drug resistance marker/GFP into the locus you're knocking out. Although the efficiency would be much lower, you'd be sure that any cells you selected actually had at least one allele knocked out. This may reduce the number of clones you need to screen to find one with all alleles knocked out.

http://www.mcu.edu.tw/department/biotec/en_page/MSR/

I used PCR followed by restriction digest to determine the genotype of my CRISPR clones. This is an elegant approach which detects both WT and KO alleles, and more importantly; makes a clear distinction between the two. 

For details check og my 2014 publication in JBC - and the extended supplement on RG. 

Cheers, Kristian

Hi there,

I think you have to go for serial dilution to have a single cell per well and then cultivate them and do the screening in case your are working with cell lines.

If you are working with ES or iPS, singularising the cells and seeding in low density allows you to isolate your clones.

You can enrich for GFP positive cells by sorting our selection by antibiotics (if your plasmid has the resistance cassete) but this will only enrich for knockout clones. You will still get wt clones.

We observed that even at a clonal population of cells, although we could see a knockout in both alleles (by PCR and Sanger sequencing), each allele showed a different INDEL mutation. So, your final and relevant result will probably be a western blot for your target protein.

Good luck

Hello Marcelo,  I also have the same question on this topic, could you please explain more on seeing the knockout in both alleles by using PCR and Sanger sequencing? I assumed that you would do a routine PCR at the gene of interest, then sequence them. but how could you tell the difference between mono-allelic or bi-allelic knockout by sanger sequencing? 

Hi Danny,

What we do here is just a single PCR of the target locus with primers annealing around 100 bp far from the DBS or nick sites. In our particular case, we have, in the wild type, a 400 bp fragment. The knock out generates a fragment of around 180 bp. In that way it is easy to distinguish wild type and mono-allelic knockouts. 

The tricky part is when we see by PCR that both alleles are knocked out but each allele displays a different repaired sequence. This can generate fragments with the same or close size and when they are sequenced you see a I mixed profile, just like two overlapped chromatograms in the region of the DSB.

But this is what we got when we used 4 guides of Cas9 nickase to generate 2 DBS and delete a region of around 100 bp.

If you are working with the Cas9 nuclease and generating only one DBS, I imagine that, as most indels do not change the fragment size very considerably, you could look carefully to your chromatogram and see if the overlapped profile of peaks corresponds to your wild type sequence which would indicate a mono allelic knockout.

By doing so, we saw that most knockouts that we got were deletions of 2 and 3 nt and so we got at the same position one peak for the wt sequence and one peak for the same sequence shifted in 2 or 3 nt.

On the other hand, this "analysis" is not possible in a bi-allelic knockout.

Hope it helped somehow.

Good luck!

Hi Marcelo, 

Thank you very much for sharing your experience! I agree; perhaps it is very difficult to analyze Cas9 nuclease knockout using routine PCR and sanger sequencing. We might consider using single molecule PCR, so we can analyze each copy separately. By any chance, does anyone have a robust protocol for single molecule PCR?

What we do is to design a forward primer exactly targeting the cutting site. If the primer binding site is disrupted, no band would be amplified after PCR. We chose the clones that without band, and do further sequencing to confirm the mutation in both allele.