CRISPR screen - enrichment vs depletion in a drug toxicity screen

[ChrisMDR]

Hi everyone, I used mouse GeCKOv2 to screen for genes which regulate/mediate toxicity of a chemotherapeutic, kind of like the first Lander/Sabitini paper in Science. My readout/phenotype was cell survival. I was hoping to find genes which, when disrupted, increase survival as well as decrease it. Using MAGeCK, I found lots of statistically significant genes in both groups. However, genes which when disrupted increase survival (i.e., these sgRNAs are enriched) are far easier to validate in independent assays than genes associated with reduced survival (i.e., these sgRNAs are depleted). For the record, the dose of drug I used was pretty harsh (equivalent to LD90). I think my problems might be due to enrichment vs depletion and that enrichment is a more reliable phenotype. Is this the case? I know people talk about drop-out screening being difficult and for sure, using an LD90 is going to favor identification of gene disruptions which lead to survival. Keen to know what people think. Thanks!

[Julia Joung]

Hi,

In your case (LD90), you see better validation for enrichment screens than for depletion screens because you are applying a very harsh selection pressure on the cells, that results in losing most of your library. Therefore, when you are looking at depleted sgRNAs, you are more likely to see sgRNAs that are depleted due to random chance because you are losing most of your cells/library. If you use LD50, which we typically screen at, then you are more likely to see depleted sgRNAs validate.

Best,

Julia

On Wed, May 1, 2019 at 11:19 AM ChrisMDR <cjmcder...@gmail.com> wrote:

Hi everyone, I used mouse GeCKOv2 to screen for genes which regulate/mediate toxicity of a chemotherapeutic, kind of like the first Lander/Sabitini paper in Science. My readout/phenotype was cell survival. I was hoping to find genes which, when disrupted, increase survival as well as decrease it. Using MAGeCK, I found lots of statistically significant genes in both groups. However, genes which when disrupted increase survival (i.e., these sgRNAs are enriched) are far easier to validate in independent assays than genes associated with reduced survival (i.e., these sgRNAs are depleted). For the record, the dose of drug I used was pretty harsh (equivalent to LD90). I think my problems might be due to enrichment vs depletion and that enrichment is a more reliable phenotype. Is this the case? I know people talk about drop-out screening being difficult and for sure, using an LD90 is going to favor identification of gene disruptions which lead to survival. Keen to know what people think. Thanks!

--
You received this message because you are subscribed to the Google Groups "Genome Engineering using CRISPR/Cas Systems" group.
To unsubscribe from this group and stop receiving emails from it, send an email to crispr+un...@googlegroups.com.
To post to this group, send email to cri...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[DavidOC]

Hi Julia/Chris, 

I wondered if you could help with a related problem. I am carrying out a positive screen using the hGECKOv2 aiming to identify genes that mediate resistance to a chemotherapeutic agent. I have been having trouble getting the drug dose right. I initially used the IC50 at 72hr and then maintained cells in this dose for 14 days, replacing drug every 72hr. However, I see a huge amount of cell death with all cells dead by around day 10. I repeated the experiment with the IC40 dose but had the same problem.

I am currently considering repeating the experiment using a lower dose for 14 days or alternatively treating cells for only 5 days and then growing for a further 9 days without drug to enrich the resistant cells. 

Any thoughts would be greatly appreciated. 

David

To unsubscribe from this group and stop receiving emails from it, send an email to cri...@googlegroups.com.

[Julia Joung]

Hi David,

I would suggest in your case to first repeat the dose curve to make sure that the IC50 value you have obtained is correct. If it is, then you should try applying the drug for 1 week and then harvesting immediately. I would not grow the cells for too long without the drug before harvesting because this will increase the noise in your screen.

A few other things to keep in mind:

1. As cells are dying in your screen, you may want to stick to a certain plating density every time you passage, i.e. after losing 80% of your cells, you scale down to smaller culture flasks/dishes. Sparse cell culture will increase your likelihood of cell death. For the vemurafenib screens, we passage and refresh drug every 2 days and scale down when necessary.

2. Select on puromycin for 5-7 days, then recover for 2 days without antibiotic before starting your drug selection. Do not do the puromycin selection and your drug selection at the same time.

Best,

Julia

To unsubscribe from this group and stop receiving emails from it, send an email to crispr+un...@googlegroups.com.

To post to this group, send email to cri...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/crispr/de5f473e-5c7d-4bee-9780-7e174e4792b4%40googlegroups.com.

[DavidOC]

Thanks Julia, that's very helpful. I've been maintaining the cells in puromycin so that may be one problem. 

I've repeated the IC50 3 times and it is always relatively similar and is in keeping with the literature so I think the dose is reasonable, although there is a very steep drop in the cell viability curves within quite a narrow dose window. 

I should have mentioned that I am using a suspension leukaemia cell line and so one issue is the accumulation of dead cells and debris. I don't know of any method to get rid of this. Have you worked with similar cell lines? 

Many thanks,

David

[Julia Joung]

I have not screened in suspension cell lines with a lot of cell death, but I think it's common to do DNAse treatment before harvesting your cells so you do not contaminate DNA from your live cells with those from the dead cells. Depending on your cell type you may also be able to spin your cells at low speed to collect only live cells. Worst case you could also sort out your live cells, but that would be a lot of work.

Best,

Julia

To unsubscribe from this group and stop receiving emails from it, send an email to crispr+un...@googlegroups.com.

To post to this group, send email to cri...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/crispr/4bc4a40e-c519-4157-909c-0c36684b50d1%40googlegroups.com.

[DavidOC]

Thanks for your advice Julia, much appreciated. 

David

[Dylan Jones]

Hi

I would like to ask - when do you add the drugs in a pooled screen.  Do you wait for cells to attach to the surface of the flask for 24hrs before adding the drug or can you add cells and drug to flasks same time?

 If your treating with drug that will kill 50% of the cells, I assume that you still have to split and count cells after 72 hrs treatment to confirm LD50,   When you re-plate cells, do you add drugs same time or wait 24hrs?

Thanks

Dylan

[Julia Joung]

Hi Dylan,

I think this depends on the drug you are using. For the vemurafenib CRISPR activation screens, during the screen we passaged the cells every other day and added fresh vemurafenib to the cells before cells have attached. The dose for LD50 was predetermined before the screen.

Best,

Julia

To unsubscribe from this group and stop receiving emails from it, send an email to crispr+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/crispr/c99687a4-4161-4b0b-af1b-0d60bc9c4793%40googlegroups.com.

[dylan...@gmail.com]

Thanks.

Have you done rescue screens with lethal dose 99% to see which sgRNA and gene when KO protect cells from the drug?  As the selection pressure is so strong, I assume you don't need a DMSO control culture for this.  My plan is to seed 40e6 cells in 4 flasks (10e6 cells/flask)  and kill 99% after 6 days of culture (400,000 cells leftover), and allow the cells to recover so that I can have enough DNA for PCR.  As I don't know how long it will take to recover,  I don't want to culture DMSO treated cells for several days as it will be expensive and a waste of time.  What do you think?

[Julia Joung]

Hi Dylan,

Generally in our experience, longer exposure to a lower dose of drug will work better than shorter exposure to a higher dose of drug. It is worth trying your approach, but if your screening results are very noisy, then try the approach I suggested.

Best,

Julia

To view this discussion on the web visit https://groups.google.com/d/msgid/crispr/b68211ae-3808-4b1f-be5f-a7f19b71c5e3n%40googlegroups.com.

[dylan...@gmail.com]

Thanks.  I have another question.  In drug screen with IC99, I have less DNA for PCR.   For the reference and control, I can run 100 x  5ug DNA in 50ul volume 24cycles, but for the drug-treated samples I have fewer cells and therefore DNA yield.  I can only run 2ug per 50ul reaction.  Would you increase the cycles in the PCR so that I would get an equivalent amount of PCR band in the gel?