A protocol for Targeted Perturb (TAP)-seq and targeted single-cell RNA-seq

Daniel Schraivogel European Molecular Biology Laboratory (EMBL), Genome Biology Unit, 69117 Heidelberg, Germany https://orcid.org/0000-0002-5734-6980 Lars Velten Center for Genomic Regulation, The Barcelona Institute of Science and Technology, Barcelona, Spain https://orcid.org/0000-0002-1233-5874 Andreas R. Gschwind Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA https://orcid.org/0000-0002-0769-6907 Lars M. Steinmetz (  larsms@embl.de ) European Molecular Biology Laboratory (EMBL), Genome Biology Unit, 69117 Heidelberg, Germany; Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA; Stanford Genome Technology Center, Palo Alto, California 94304, USA https://orcid.org/0000-00023962-2865


Introduction
Single-cell RNA-sequencing (scRNA-seq) is revolutionizing basic and applied biomedical research. Among diverse applications, scRNA-seq readouts of pooled CRISPR genetic screens, also referred to as CROP-seq or Perturb-seq, enabled to identify the consequences of de ned genetic perturbations on gene expression and cellular differentiation (Adamson et al. 2016, Dixit et al. 2016, Jaitin et al. 2016, Datliner et al. 2017. Although this concept has captured widespread attention, its broad applicability has been prohibited due to (1) high sequencing costs, (2) bottlenecks in data analysis, and (3) a lack of sensitivity of scRNA-seq methods.
Here we overcome these limitations by introducing a protocol for targeted Perturb-seq (TAP-seq) ( Figure  1). The method has been established primarily as a readout for pooled CRISPR screens. However, TAPseq is applicable to diverse single-cell applications in mammalian cell lines and primary mammalian cells, with or without CRISPR perturbations. Centrifuges compatible with 10X Genomics, e.g. 1 million cells per ml. Double-check cell numbers and absence of multiplets by cell counting.

Reagents
b. If low numbers of cells are available: Sort cells directly into 20 µl cell collection buffer provided in a 1.5 ml tube. Cell numbers are deduced from ow cytometry counts, and by adding a correction factor of 1 .5x to 2x (has to be adapted, start with 1.5x). Cells are not counted again.
Example: The correction factor considers cell loss during handling of cells after sorting, and the dilution factor during 10X Genomics single-cell Master Mix preparation and Chromium run. If 250 cells should be sequenced (and a correction factor 1.5x), sort 250 x 1.5 = 375 cells into 20 µl cell collection buffer. Approximately 3.5 nl (100 µm nozzle) or 7 nl (130 µm nozzle) liquid will be added on top per sorted cell. After sort is nished, vortex quickly to collect cells from tube wall, spin cells shortly, and measure volume. Then ll up with collection buffer to 33.8 µl, which is the volume added into the 10X Genomics single-cell master mix.
Comment: We generally use 2.5 µl of outer primer mix for a panel size < 500 different primers, but amount can be increased to 4 µl per reaction in case more primers are present in panel.
Comment: For 10X v2, v3 and NextGEM, no adaption for PCR1 and PCR2 is necessary, since the Partial Read 1 primer can be used for all three protocols.
Comment: The CROP-seq vector primer (CROPouter, CROPinner) can be omitted in case no CROP-seq vector is present, or replaced if sgRNA/barcode is expressed by a vector system other than CROP-seq.
Optional: 5 µl of the 10X Genomics cDNA from step 15 can be put on side for separate CROP-seq vector ampli cation. Follow the protocol as described from here on, but omitting the outer and inner primer mix. Comment: Exact cycle number has to be determined, but we achieved good results with 11 to 14 cycles. The cycle number depends on the number of cells, the amount of input cDNA, the number of ampli ed genes, and the expression levels of these genes. As a start, we suggest 12 cycles, which should result in ³ 10 ng of PCR product. Too high cycle numbers of PCR1-3 might result in a high MW product after PCR3 peaking around 1,000 to 1,500 bp; we generally do not observe any issues from this high MW product after Illumina sequencing and data analysis (product might not cluster e ciently during Illumina run preparation).

Anticipated Results
Typical Bioanalyzer traces for TAP-seq are shown in Figure 3.