Screening for Genes Enabling Robust Growth

The screens for genes required for robust growth were conducted essentially as previously described (Gilbert et al., 2014 (link)). Briefly, plasmid libraries were packaged into lentivirus in HEK293T cells (RRID:CVCL_0063) and infected into a previously established polyclonal K562 cell line stably expressing dCas9-KRAB grown in 3L spinner flasks (Bellco, Vineland, NJ). After two days, infected cells were selected with 0.75 μg/mL puromycin (Tocris, Bristol, UK) for two days, allowed to recover for one day, and then cultured at a minimum of 750 × 10⁶ cells in 1.5L standard media (RPMI-1640 with 10% Fetal Bovine Serum and 1x supplemental glutamine, penicillin, and streptomycin) from 'T0' to 'endpoint,' determined by ~10 cell doublings after T0. CRISPRi screen cells were mock-treated with 0.1% DMSO (Sigma-Aldrich, St. Louis, MO) but otherwise left untreated. Screens were performed as independent replicates starting from the infection step. The K562 dCas9-KRAB and SunTag-VP64 cell lines were obtained from (Gilbert et al., 2014 (link)) and had been constructed from K562 cells originally obtained from ATCC (RRID:CVCL_0004). Cytogenetic profiling by array comparative genomic hybridization (not shown) closely matched previous characterizations of the K562 cell line (Naumann et al., 2001 (link)). All cell lines tested negative for mycoplasma contamination (MycoAlert Kit, Lonza, Basel, Switzerland) in regular screenings.
Frozen samples of 250 × 10⁶ cells collected at T0 and endpoint were processed as previously described (Gilbert et al., 2014 (link)), with the substitution of an SbfI restriction digest (SbfI-HF, New England Biolabs, Ipswich, MA) in place of the MfeI digest in the genomic DNA fragmentation and enrichment step. The sgRNA-encoding regions were sequenced on an Illumina HiSeq-4000 using custom primers. Sequencing reads were aligned to the expected library sequences using Bowtie (v1.0.0, [Ben Langmead et al., 2009 (link)]) and read counts were processed using custom Python scripts (available at https://github.com/mhorlbeck/ScreenProcessing) based on previously established shRNA screen analysis pipelines (Bassik et al., 2013 (link); Kampmann et al., 2013 (link)). sgRNAs represented with fewer than 50 sequencing reads in both T0 and Endpoint were excluded from analysis. sgRNA growth phenotypes (γ) were calculated by normalizing sgRNA log₂ enrichment from T0 to endpoint samples and normalizing by the number of cell doublings in this time period. CRISPRi v1 screen data from Gilbert et al. (2014) (link) was re-analyzed using this pipeline, and the hCRISPRi/a-v2 5 sgRNA/gene libraries were evaluated by analyzing the sgRNA read counts corresponding to only the 5 sgRNA/gene sublibraries. Gene ontology analysis was conducted using DAVID 6.7 (Huang et al., 2009 (link)) with default search categories and with background lists representing the genes targeted by the CRISPRa v1 or hCRISPRa-v2 libraries where appropriate. For Figure 4B and Figure 4—figure supplement 1D, 'shared hit' genes were 70 genes that scored as strong anti-growth hits (phenotype z-score x –log₁₀ p-value ≤ −10) in both CRISPRa v1 and hCRISPRa-v2.