Uncoupling of sgRNAs from their associated barcodes during PCR amplification of combinatorial CRISPR screens

Abstract

Many implementations of pooled screens in mammalian cells rely on linking an element of interest to a barcode, with the latter subsequently quantitated by next generation sequencing. However, substantial uncoupling between these paired elements during lentiviral production has been reported, especially as the distance between elements increases. We detail that PCR amplification is another major source of uncoupling, and becomes more pronounced with increased amounts of DNA template molecules and PCR cycles. To lessen uncoupling in systems that use paired elements for detection, we recommend minimizing the distance between elements, using low and equal template DNA inputs for plasmid and genomic DNA during PCR, and minimizing the number of PCR cycles. We also present a vector design for conducting combinatorial CRISPR screens that enables accurate barcode-based detection with a single short sequencing read and minimal uncoupling.

Fig 1. Schematic of oligonucleotide design and combinatorial library cloning strategy proposed in this study.

BC: barcode.

Fig 2. Comparison of uncoupling across sample types, input amounts, and number of PCR cycles.

(A) Uncoupling between SaCas9 sgRNAs and SpCas9 sgRNAs under various PCR conditions. Each box represents 57 paired sgRNAs, plotting the fraction of reads for which the sgRNAs were correctly paired. The line represents the median, the box the 25th and 75th percentiles, and the whiskers the 10th and 90th percentiles. Our initial PCR conditions (28 cycles with 10 ng pDNA and 10 μg gDNA) led to substantial uncoupling. (B) Uncoupling between sgRNAs and their associated barcodes under various PCR conditions. (C) Uncoupling between barcodes under various PCR conditions. Box and whisker plots in (B) and (C) are the same as in (A).

Fig 3. Comparison of uncoupling at various distances between linked elements.

Data are replotted from Fig 2. Uncoupling between barcodes (separated by 17 nts), an sgRNA and its associated barcode (82 nts), and sgRNAs (193 nts). PCR was performed at either 16, 22, or 28 cycles using 10 pg pDNA (left) or 10 μg gDNA (right).

Fig 4. Comparison of PCR polymerases with pDNA and gDNA.

(A) Uncoupling between sgRNAs, using 28 cycles of PCR and a range of pDNA inputs. (B) Uncoupling between sgRNAs, using 28 cycles of PCR and either 1 or 10 μg of gDNA. Only Ex Taq and LA Taq produced a product with 10 μg of gDNA input.

Fig 5. Gel image of PCR products comparing polymerase performance with 1 μg of gDNA input.

Product bands are 650nt and indicated with yellow arrows. Blue arrows indicate unused primer bands. Both images are of the same 2% Agarose gel at two different exposures. In both exposures, LA Taq, Ex Taq, KOD and NEB Next produce robust product bands. Herculase and Fusion produce very faint product bands only visible with overexposure. Q5 produced no visible band even with overexposure. The ladder is the 1KB Plus (ThermoFisher, 10787018).

Fig 6. Schematics of various vector designs used for combinatorial CRISPR screens.

See also Table 1. BC: barcode. RE: restriction enzyme.