Pooling WT and CRISPR/Cas9 KOs cells in the same scRNAseq sample and then demultiplex by puromycin resistance cassette expression

Question

[Francesco]

Library preparation for single cells RNA sequencing ends being the most expensive part of the of the process, so it's way cheaper doubling the reads depth than the number of samples. I have seed already somebody pooling mouse and human cells in the same sample and then demultiplex the cells by using unique transcripts to reduce the costs of scRNAseq. Also, I just ended up reading of the so-called CROP sequencing in which you use the sgRNA transcript to demultiplex multiple cells with different sgRNA vectors.

I wish to single cell sequence a pooled sample of WT cells and CRISPR KOs and demultiplex them later. Unfortunately, the expression of my target gene is not sky high so it's likely I'll have lots of dropouts for that mRNA and I cannot use the CAS9 induced InDels to distinguish WT from KOs. Also, the plasmid I used was not designed for CROP seq so I don't have any reporter sgRNA with a poly A tail. I was thinking instead of using the puromycin resistance cassette to separate the KOs from the WT since it should be stably expressed by the EF-1α promoter.

Does anyone have any experience or comments on that practice?

Answer 1

[jockbanan]

Hi!

If you have just a few samples that you want to pool and process together and demultiplex later, a good and frequently used solution is cell hashing

https://cite-seq.com/cell-hashing/

About the use of CROP-seq cassette, if you have established CROP-seq, you can actually use the cassette to label cells by infecting them with viruses carrying non-targeting labels, instead of functioning sgRNAs, so that these won't do anything even in Cas9-expressing cells. Thanks to the CROP-seq cassette structure, the cells will then express the label as a transcript with a polyA tail and thanks to that, the labels will be visible in the scRNA-seq data. But this is a more complicated setup and if you don't pool dozens or hundreds of samples, it is much easier to use cell hashing.