Tide analysis on single clones - help on results

[Micky.boniotto]

Good morning

I'm trying to isolate a single clone modified by CRISPR/Cas9
I used the pSpCas9(BB)-2A-GFP (PX458) to transfect my cells. I isolated the GFP+ cells by cell sorter and finally I collected hundred 96-wells of clones growing from a single GFP+ cell.
I run a PCR of about 750 bp from the DNA isolated from the clones to identify clones that were inactivated for both alleles (I need in fact a KO).
The plan is to identify a possible KO by TIDE as this program can estimate the percentage of two alleles when the Cas9 has cut and the NHEJ repair has created different alleles. First question : anyone has already used sequencing and TIDE analysis to identify the KO clones?

In a preliminary analysis of 4 clones, 3 clones had a deletion/insertion on at least one allele and 1 clone was WT (the sequence aligns with the reference sequence).
So when I analyse the 3 clones against the WT sequence by TIDE, I obtain the results depicted in the 3 figures attached.
I would like to know whether I can estimate if these 3 clones are actually KO or I need to clone the PCR products and sequence the different clones or I need to run a WB to see whether the protein is still expressed.

I would really like to know your ideas!

Cheers

 

[YANG]

I think your #1 AND #3 are KO. #2 maybe mixed with another colony with larger than 10 bp indels. You still need to do a WB to confirm.

[Micky.boniotto]

Another question.
After CRISPR/Cas9 with Px458, enrichment by sorter of GFP+ high cells, do you expect many KO clones?
I just checked other clones by sequencing and it seems that all of them have both alleles modified after Cas9 cut. Do you expect always to have both alleles cut by the Cas9? or do you expect the majority to have ony 1 allele that is cut and rearranged after NHEJ?
Cheers for your reply

 

[YANG]

It depends on how efficient is your guide RNA. Based on your description, looks like it's highly efficient. So if you need heterozygous ko clone, you need to lower the concentration of Cas9 construct.

[Fatwa Adikusuma]

Yes, we normally get close to 100% efficiency after selection particularly when using drug selection (puro). 

Cheers,

Fatwa

[abhishek...@synthego.com]

I would suggest using ICE analysis from Synthego (ice.synthego.com). Its a free software that analyzes your Sanger data and will give you %indels & indel contributions that are going to be more informative regarding identification of KO clones, that are accurate with NGS. 

The data you show suggests that you have a triploid cell line for this gene, i.e., three alleles. It is going to be very difficult to determine KO using the data above, but might be easier with the analysis through ICE where you can see indel contributions.  We have effectively used the ICE software to identify single cell clones that are genetic knockouts. 

[Phil Abbosh]

guys, we are doing something similar to what is posted in this thread and have a bit of a head scratcher for a result. We had a result that by both TIDE and ICE said there was ~60-70% editing efficiency. Because this was on a single cell clone, we suspected that we had multiple cells in our original well and so subcloned that well into single cells again.  For all 12 subclones, we had the exact same TIDE/ICE result though.

We have since had similar findings with single cell clones but had 80/20 or 75/25 edits etc.  we were kind of expecting a 0/50/100 output, not something in between. Seems like either:

TIDE/ICE arent very accurate with the indel %

Cells have >2 copies of the genes we are trying to edit

We got super duper unlucky with our subcloning of a clone

Next step is probably an amplicon NGS. Anyone with similar experience, ideas of what is happening, or suggestions?

[Keith Breau]

What cell line are you attempting to edit? This sounds kind of like polyploidy. I know HEK lines, for example, are triploid at a lot of loci.

[Phil Abbosh]

Hi Keith

good suggestion but pretty sure that's not it.  We are editing organoids from a normal tissue we took out of a mouse.  this has happened with multiple genes now.  It's possible that in the process of culturing them, they acquired CNAs but seems unlikely to have happened at all the genes we are interested in. Its also strange that sometimes our TIDE/ICE result would suggest there are 5 alleles, sometimes 3 etc.

We did KO p53 and Rb1 first in some of the organoids, then sprinkle in addtional genes, so it might be more likely in those, but again, we started from 'normal' tissue. we are going to be doing whole exome sequencing on these soon, so we will know if there are indeed CNAs.  I am going to assume for now there arent though.

[Keith Breau]

What does TIDE call as the predominant alleles? A modest size deletion would be amplified better during PCR and be over-represented. Alternatively there are polymerase pause sites throughout the genome, if you delete a few bases in one of these it could also skew PCR. Though the fact that you are seeing this at many sites suggests a more systemic problem. Can you explain a little more about your experiment? p53 KO can definitely induce polyploidy, especially if your gRNA has poor specificity and induces DSBs around the genome. How many sites did you target, on what chromosomes, and how many are showing this result? Did you run your gRNAs through any program to validate specificity?

[Gavin Kurgan]

In addition to Keith’s comments, it might be good to look if your WT sequence or the sequence of the predicted allele from sequencing have problematic amplification/sequencing motifs (e.g homopolymers, predicted secondary structures, etc.) which may be causing bias. Sanger deconvolution based methods definitely have limitations on accuracy as well, although it may be hard to say if that is responsible without NGS or some other assay.

[Phil Abbosh]

thanks a lot Keith/Gavin. let me reply to both of you:

Keith: for the most recent sample, we have insertion of a single A at the cut site see below. The top chromatogram is the edited, bottom is WT

I agree, a moderate indel might cause skewed amplification but I think that is less likley than with a 1 bp insertion. 

the guide targets mouse chromosome 16, gene Ercc4.  I dont routinely run the guides through an off target program because the one I use weighs that in to the output file that it gives me (https://portals.broadinstitute.org/gpp/public/analysis-tools/sgrna-design). this crRNA happened to be the 5th best when accounting for predicted on target PLUS off target activity. of the ~400 guides that the site gave me, this one was ranked ~22nd for off target score.  I honestly dont know if that is good or bad, but it would be in the top ~5% of guides for that gene, so I suppose it should have less off target effects than others I could have chosen.  Again, I dont know exactly what 22nd place or top 5% means.  Possible that all the other guides were not very good either and the whole bunch is bad.

Gavin: There is a homopolymer at the 5' end of the amplicon and we didnt appreciate that til after we sequenced it.  you will notice in the trace that there is a little bit of uncleanliness and I think it is bc of that upstream homopolymer. we sequence the amplicon with the R primer and the baseline is cleaner but we still get the same result.

After looking at the traces from our other genes a little more closely, I think you are right Gavin, that these arent a replacement for NGS or TA cloning of a pool of amplicons.  i think that is where this will go next.  Maybe this is closer to 50/50 and ICE/TIDE overestimate the indel %.

thanks for your help guys!

[Keith Breau]

Hi Phil. Seeing those data and you response, I think your problem is one of sequencing noise. I never use one of the PCR amplification primers for sequencing as this tends to generate a lot of excess noise that throws off TIDE results. I would highly recommend designing a third primer that sits inside your amplicon, just inside one of the PCR primers. If you look, the noise in your WT sample matches your newly-predominant indel in your edited cells perfectly. 

On Tuesday, January 30, 2018 at 5:55:30 AM UTC-5, Micky.boniotto wrote:

[Diego V]

In your WT sample the background is a lag in one base (the small peak behind the next base is the same as the previous one) and also in the edited sequence. If you say that there is a 5' homopolymer, that could explain these data.

Macrogen also explain these results in his troubleshooting guide as a slippage mediated by an homopolymer

https://dna.macrogen-europe.com/eng/support/ces/guide/troubleshooting.jsp

[Phil Abbosh]

Wow, great catch!

--
You received this message because you are subscribed to a topic in the Google Groups "Genome Engineering using CRISPR/Cas Systems" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/crispr/SzdWw8NQWzs/unsubscribe.
To unsubscribe from this group and all its topics, send an email to crispr+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/crispr/a7d40c27-8a79-4450-9cb2-76906ae11176o%40googlegroups.com.