(PDF) Circular synthesized CRISPR/Cas gRNAs for functional interrogations in the coding and noncoding genomeArticlePDF AvailableCircular synthesized CRISPR/Cas gRNAs for functional interrogations in the coding and noncoding genomeMarch 2019eLife Sciences 8DOI:10.7554/eLife.42549Project: Mapping the Landscape of Genetic Dependencies in CancerAuthors: Martin WegnerGoethe-Universität Frankfurt am Main Valentina DiehlGoethe-Universität Frankfurt am Main Verena BittlVerena BittlThis person is not on ResearchGate, or hasn t claimed this research yet. Rahel de BruynGoethe-Universität Frankfurt am MainShow all 15 authorsHide Download full-text PDFRead full-textDownload full-text PDFRead full-textDownload citation Copy link Link copied Read full-text Download citation Copy link Link copiedCitations (14)References (87)AbstractCurrent technologies to generate CRISPR/Cas gene perturbation reagents are labor intense and require multiple ligation and cloning steps. Furthermore, increasing gRNA sequence diversity negatively affects gRNA distribution, leading to libraries of heterogeneous quality. Here, we present a rapid and cloning-free mutagenesis technology to efficiently generate covalently-closed-circular-synthesized (3Cs) CRISPR/Cas gRNA reagents that uncouples sequence diversity from sequence distribution. We demonstrate fidelity and performance of 3Cs reagents by tailored targeting of all human deubiquitinating enzymes (DUBs) and identify their essentiality for cell fitness. To explore high-content screening, we aimed at generating the up-to-date largest gRNA library to simultaneously interrogate the coding and noncoding human genome and identify genes, predicted promoter flanking regions, transcription factor and CTCF binding sites linked to doxorubicin resistance. Our 3Cs technology enables fast and robust generation of bias-free gene perturbation libraries with yet unmatched diversities and should be considered an alternative to established technologies. Discover the world s research20+ million members135+ million publications700k+ research projectsJoin for freePublic Full-text 1Content uploaded by Manuel KaulichAuthor contentAll content in this area was uploaded by Manuel Kaulich on Mar 19, 2019 Content may be subject to copyright. *For correspondence:kaulich@em.uni-frankfurt.de†These authors contributedequally to this workCompeting interest: Seepage 26Funding: See page 27Received: 07 October 2018Accepted: 25 February 2019Published: 06 March 2019Reviewing editor: Jonathan SWeissman, University ofCalifornia, San Francisco, UnitedStatesCopyright Wegner et al. Thisarticle is distributed under theterms of the Creative CommonsAttribution License, whichpermits unrestricted use andredistribution provided that theoriginal author and source arecredited.Circular synthesized CRISPR/Cas gRNAsfor functional interrogations in the codingand noncoding genomeMartin Wegner1†, Valentina Diehl1†, Verena Bittl1,2, Rahel de Bruyn1,Svenja Wiechmann1,3, Yves Matthess1, Marie Hebel1, Michael GB Hayes4,Simone Schaubeck1, Christopher Benner4, Sven Heinz4, Anja Bremm1,2,Ivan Dikic1,2,5,6, Andreas Ernst1,3, Manuel Kaulich1,5,6*1Institute of Biochemistry II, Goethe University Frankfurt - Medical Faculty,University Hospital, Frankfurt, Germany;2Buchmann Institute for Molecular LifeSciences, Goethe University, Frankfurt, Germany;3Project Group TranslationalMedicine Pharmacology TMP, Fraunhofer Institute for Molecular Biology andApplied Ecology IME, Frankfurt, Germany;4Department of Medicine, University ofCalifornia, San Diego, San Diego, United States;5Frankfurt Cancer Institute,Frankfurt am Main, Germany;6Cardio-Pulmonary Institute, Frankfurt am Main,GermanyAbstract Current technologies used to generate CRISPR/Cas gene perturbation reagents arelabor intense and require multiple ligation and cloning steps. Furthermore, increasing gRNAsequence diversity negatively affects gRNA distribution, leading to libraries of heterogeneousquality. Here, we present a rapid and cloning-free mutagenesis technology that can efficientlygenerate covalently-closed-circular-synthesized (3Cs) CRISPR/Cas gRNA reagents and thatuncouples sequence diversity from sequence distribution. We demonstrate the fidelity andperformance of 3Cs reagents by tailored targeting of all human deubiquitinating enzymes (DUBs)and identify their essentiality for cell fitness. To explore high-content screening, we aimed togenerate the largest up-to-date gRNA library that can be used to interrogate the coding andnoncoding human genome and simultaneously to identify genes, predicted promoter flankingregions, transcription factors and CTCF binding sites that are linked to doxorubicin resistance. Our3Cs technology enables fast and robust generation of bias-free gene perturbation libraries with yetunmatched diversities and should be considered an alternative to established technologies.DOI: https://doi.org/10.7554/eLife.42549.001IntroductionCRISPR/Cas has rapidly become the gold standard for unbiased high-throughput experiments, out-performing preexisting technologies such as RNAi (Evers et al., 2016;Morgens et al., 2016). A fun-damentally important aspect of high-fidelity CRISPR/Cas screening is the quality of the gRNA librarythat is interrogated, with its diversity and distribution primarily influencing downstream experimentalscales (Sanson et al., 2018). Conventionally used methods to generate gRNA libraries in pooled orarrayed formats include T4 ligase or homology-based cloning techniques, which require the PCR-based amplification of gRNA-encoding oligonucleotides as well as the presence of open plasmidDNA for successful gRNA sequence cloning (Arakawa, 2016;Koike-Yusa et al., 2014;Ong et al.,2017;Schmidt et al., 2015;Shalem et al., 2014;Vidigal and Ventura, 2015;Wang et al., 2014).Owing to these technical constraints, conventional libraries contain an unwanted PCR and cloning-dependent bias in their gRNA distribution that influences the experimental scale required forWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 1 of 31TOOLS AND RESOURCES statistically significant hit calling (Shalem et al., 2014;Wang et al., 2014). CRISPR libraries havebecome ubiquitously used in functional genomics efforts, underscoring relevance and utility of newPCR- and cloning-free technologies.The rod-shaped filamentous phage M13 differs from other bacteriophages in that its genome-packaging capacity is variable and in that it is present as single-stranded (ss) DNA. Kunkel mutagen-esis utilizes M13’s malleable coat and the ease of ssDNA purification from M13 phage and enablesrapid site-specific mutagenesis to construct high-quality phage display libraries (Handa and Var-shney, 1998;Huang et al., 2012;Kunkel, 1985;Kunkel, 2001). Kunkel mutagenesis has signifi-cantly contributed to the great success of phage display technologies (Ernst et al., 2013;Sidhu, 2001).Here, we demonstrate the applicability of Kunkel mutagenesis in the generation of high-qualityand high-fidelity CRISPR/Cas and RNAi gene perturbation reagents. In more detail, we developed ahighly reproducible improved Kunkel mutagenesis technology that is designed to generate 3CsCRISPR/Cas gRNA libraries robustly over a broad range of gRNA diversities. We demonstrate thehigh fidelity of 3Cs gRNA libraries by targeting all human DUBs and then determining their prolifer-ative depletion phenotype, confirming previously known and discovering hitherto unknown DUBphenotypes. In an effort to enable unbiased screening within coding and noncoding regions, weencoded SpCas9 nucleotide preferences into a degenerated oligonucleotide and generated a highlycomplex CRISPR/Cas gRNA library (truly genome-wide (TGW)). Doxorubicin-positive selectionscreens with the TGW library in unperturbed human telomerase-immortalized retinal pigmented epi-thelial cells (hTERT-RPE1) were used to identify coding and noncoding regions, emphasizing the rel-evance of noncoding sequence elements in drug-resistance mechanisms. To enable high-contentfunctional interrogations on a truly genome-wide scale, we introduce an optimized version of thislibrary (oTGW). In summary, we establish the 3Cs technology as a robust alternative method for thegeneration of high-quality CRISPR/Cas gene perturbation libraries.ResultsCircular synthesized gRNAs are high-quality CRISPR/Cas reagentsIn classical Kunkel mutagenesis (Kunkel, 1985;Kunkel, 2001), the circular ssDNA isolated from fila-mentous phage is hybridized with a complementary oligonucleotide that is extended and ligated toobtain a double-stranded DNA plasmid. As Kunkel mutagenesis is performed on ssDNA, we antici-pated that it would be insensitive to the secondary DNA structures of viral sequence elements andtherefore should enable the PCR and cloning-free generation of lentiviral gene perturbationreagents (Huang et al., 2012;Kunkel, 1985). We therefore hypothesized that the generation of len-tiviral CRISPR/Cas gRNA libraries using circular ssDNA and Kunkel mutagenesis would reducethe coupling of gRNA diversity to gRNA distribution and would generate reagents of high quality(Figure 1A).To demonstrate its general applicability to lentiviral CRISPR/Cas plasmids, we transformedEscherichia coli CJ236 bacteria with the commonly used pLentiGuide and pLentiCRISPRv2 plasmids(Sanjana et al., 2014), both of which contain a U6 promoter-controlled non-human targeting (NHT)placeholder gRNA followed by a SpCas9 tracrRNA sequence (Doench et al., 2014;Sanjana et al.,2014). Importantly, F-factor-containing CJ236 bacteria lack dUTPase (dut–) and uracil-glycosylase(ung–), and therefore tolerate the presence of deoxyuridine (dU) in genomic and plasmid DNA(Kim and Wilson, 2012). Superinfection of single-colony CJ236 culture with M13KO7 bacteriophage(108cfu/mL) facilitated the generation of 30 mg of dU-containing circular ssDNA. Although circularssDNA is identical in length to dsDNA, the circular ssDNA of lentiviral CRISPR/Cas plasmidsmigrated faster and appeared as a single band in gel electrophoresis (Figure 1B). Circular dU-ssDNA was hybridized with a gRNA-encoding complementary oligonucleotide that containedsequence homology regions (3Cs homology) at its 50and 30ends, and then extended and ligatedwith T7 polymerase and T4 ligase, respectively (Figure 1A). This resulted in heteroduplexed 3CsDNA (3Cs-dsDNA), which were composed of dU-template ssDNA and deoxythymidine-containingnewly synthesized complementary DNA that also includes the gRNA-encoding oligonucleotide(Figure 1A) (Huang et al., 2012;Kunkel, 1985;Kunkel, 2001). To gain insights intothe oligonucleotide requirements and kinetics of 3Cs reactions, we tested different 3Cs homologyWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 2 of 31Tools and resources Cell Biology Genetics and Genomics breakpoint#1AB CssDNA3Cs-dsDNAkb1210.524++++GuideCRISPRv2pLentiGFP #1-6 gRNA EF H100806040200% GFP-positive cellsNHT GFP#1-6(lentiviral sup.)dsDNAssDNAkb1210.524++++Guide CRISPRv2pLenti++pcDNA302468gRNA pair copy number (log10)GFP gRNA(#)1 2 3 4 5 6GFP33.18%G10.52-NHTGFP#1GFP#1-6+ T7 Endonuclease Ikb11 .73.27.66.34Figure 1, Wegner et al.IGFP-NHTGFP#1GFP#1-6Tub020406080100% of SequencesD0123-1gRNA pair copy number(reads to median normalized, log10)1kb Plusamplification oflibrary backbonein CJ236purification ofdU-ssDNA,oligonucleotide annealingdU dUdU+ helper phageM13K07oligonucleotideextensionwith T7 DNA Pol.heteroduplexdU-CCC-dsDNA(3Cs)generationdU dUdUelectroporation purification offinal gRNAlibraryf1 origRNA gRNAf1 oridU dUdUgRNA expression cassette f1-origin (phagemid) new gRNA Oligonucleotide(s) dU mismatchWTnewf1-origin plasmid intodut-/ung- bacteriapick colony,add M13K07(108)(Kana after 2hrs)expandculturepurify phagesand ssDNAssDNA to PO4-oligos,add T7 DNA polymerasepurify dsDNA,perform QCpurify 3Cs-DNA,electroporate intodut+/ung+ bacteriao/n 6hrs 20hrs 4hrs 2hrs to o/n 4hrso/ntotal vs. hands-on time: 2d / 1hr 1d / 5hr 1d / 4hr 30ug10ng 2ug:60ng 3ug 250ugUridine: +-NHTNHTGFP #1-6GFP #1-6GFP #1-6I-SceI: +-breakpoint#2Figure 1. The 3Cs technology - covalently-closed-circular-synthesized (3Cs) CRISPR/Cas gRNA reagents. (A) The general 3Cs workflow. The individualsteps of the protocol (grey arrows), time requirements (on top of arrow) and used or expected DNA yields (below arrow) are highlighted. Timerequirements are separated by total versus hands-on time (grey scaled bars). Please note that the protocol contains two possible break points (red stopsigns) at which purified phages can be stored at 4˚C (break point #1) or bacterial pellets/purified plasmid DNA can be stored at 20˚C (break point #2).Figure 1 continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 3 of 31Tools and resources Cell Biology Genetics and Genomics lengths of 10, 13, 15, and 18 nucleotides, performed a 3Cs reaction time series, anddemonstrated that 18 nucleotides of homology (above 45˚C annealing temperature) and 2 hr of 3Csreaction time were sufficient (Figure 1—figure supplement 1A–C) (Kunkel, 2001).Using rule set 2 (RS2) (Doench et al., 2014,Doench et al., 2016), we designed six GFP-targetinggRNA sequences and extended them by 50and 303Cs homology. Synthesized gRNA-encoding oli-gonucleotides were hand-pooled in equimolar ratios, phosphorylated and used in a 1:5 ratio (2 mgssDNA to 60 ng oligonucleotide) to generate heteroduplex dU-3Cs-sDNA (Figure 1C). To removeNHT/dU-containing template and to amplify the gRNA-encoding complementary strand, 3Cs prod-ucts were column-purified and transformed in dut+/ung+bacteria. Bacterial clones were grown andtheir plasmid DNA Sanger sequenced, revealing that 81% of pLentiGuide and 82% of plenti-CRISPRv2 contained GFP-targeting gRNAs (Figure 1D and Figure 1—figure supplement 1D). Totest whether dU supplementation reduces the amount of NHT-containing template plasmid byimproving dU-incorporation during ssDNA production, CJ236 culture medium was supplementedwith 2.5 mM dU. In addition, the gRNA placeholder sequence of pLentiGuide and plentiCRISPRv2was changed to contain an I-SceI restriction enzyme recognition site. Although the effect ofincreased dU concentrations was negligible, I-SceI-mediated removal of wildtype plasmids reducedtheir level to below our detection limit (Figure 1D and Figure 1—figure supplement 1D–F). Weperformed next-generation sequencing (NGS) on the plentiCRISPRv2 sample with an average readcount of 1.15 million per GFP sequence and identified a wildtype rate of below 0.3% in the absenceof any apparent sequence bias (Figure 1E and Supplementary file 1). A one-sided Chi-squared testfor goodness of fit identified a uniform distribution (p=0.1) of all six gRNA sequences. The uniformgRNA distribution was supported by a low coefficient of variation (CV) of 33.18% and an area underthe curve (AUC, Lorenz curve) of only 0.56 (Figure 1E–F and Figure 1—figure supplement 1G).To test for the cellular functionality of 3Cs gRNAs, we used the plentiCRISPRv2 GFP-targeting3Cs gRNA constructs to generate infectious lentiviral particles and transduced GFP-positive humantelomerase-immortalized retinal pigmented epithelial (hTERT–RPE1) cells. Seven days post-transduc-tion, we performed a T7 Endonuclease I assay and observed robust GFP gene editing, both by a sin-gle GFP-targeting 3Cs gRNA (3Cs-gRNA) and by the pool of six 3Cs-gRNAs, whereas un-transducedFigure 1 continuedIn more detail, f1-origin containing double-stranded CRISPR/Cas plasmids are converted to dU-containing circular ssDNA. Guide RNA sequence(orange triangle) containing oligonucleotides (orange arrows) are annealed to ssDNA, and extended and ligated by T7 DNA polymerase and T4ligase, respectively. Heteroduplex dU-3Cs-DNA is transformed into base-excision-repair-sufficient bacteria to deplete template DNA (grey strand) andto amplify the newly synthesized DNA (orange) selectively. (B) Lentiviral CRISPR/Cas plasmids (pLentiGuide, pLentiCRISPRv2) and the mammalian cDNAexpression plasmid pcDNA3 (positive control) were converted to dU-containing circular ssDNA and analyzed by gel electrophoresis. Although identicalin size, circular ssDNA appears as a single band and migrates faster than the corresponding dsDNA form. (C) The lentiviral circular ssDNA of panel (B)was annealed with a pool of six oligonucleotides, encoding six GFP-targeting gRNAs, to generate a pool of 3Cs-dsDNA and analyzed by gelelectrophoresis. A successful 3Cs in vitro reaction is indicated by three distinct 3Cs-dsDNA product bands (Huang et al., 2012). (D) Bar graph showingthe degree of template remnants in the final 3Cs products in the presence and absence of additional Uridine in the phage culture medium as well as anI-SceI clean-up step. The gRNA libraries from panel (C) were sequenced by NGS before and after I-SceI restriction enzyme digest. Although the effectof Uridine is marginal, an enzymatic digest with I-SceI removes template plasmid remnants. (E, F) gRNA distribution displayed as raw read count datapoints (E) and normalized values in box plot format (F). The coefficient of variation was calculated by dividing the standard deviation by the mean of thelibrary’s read counts and is displayed as percentage above the box plot (F). Data were derived from the NGS data shown in panel (D). The final GFP-targeting 3Cs-gRNA library is free of sequence bias, as demonstrated by the low coefficient of variation of 33.18%, and by the uniform sequencedistribution ((E), also see Figure 1—figure supplement 1H). (G) GFP-expressing hTERT–RPE1 cells were transduced with lentiviral 3Cs-gRNA constructs(non-targeting control gRNA (non-human target sequences (NHT)), a single GFP-targeting 3Cs-gRNA (GFP#1) or a pool of six GFP-targeting 3Cs-gRNAs (GFP#1–6)), and selected with puromycin before GFP gene editing was analyzed by T7 endonuclease I assay (Guschin et al., 2010). Individualband intensities were quantified (black numbers). An empty control (–) served as the reference. (H) A dose-dependent reduction of GFP fluorescencewas determined by the flow cytometry of GFP-expressing hTERT–RPE1 cells and transduced with increasing volumes of lentiviral supernatant containinga pool of six GFP-targeting 3Cs-gRNAs (GFP#1–6). Error bars represent standard deviations (SDs) over three biological replicates (n= 3). (I) Immunoblotanalysis of hTERT–REP1 cells treated as in panel (G) demonstrates that GFP-targeting 3Cs-gRNAs induce a 3- to 4-fold reduction in total GFP proteinlevels over three biological replicates (n= 3, for quantification see also Figure 1—figure supplement 1I).DOI: https://doi.org/10.7554/eLife.42549.002The following figure supplement is available for figure 1:Figure supplement 1. Determining 3Cs parameters, I-SceI template remnant removal, and the GFP library.DOI: https://doi.org/10.7554/eLife.42549.003Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 4 of 31Tools and resources Cell Biology Genetics and Genomics (–) and an NHT control gRNA failed to edit the GFP gene (Figure 1G). GFP gene editing translatedto a lentiviral dose-dependent loss of GFP protein when analyzed by fluorescence flow-cytometryand immunoblotting (Figure 1H–I and Figure 1—figure supplement 1H). Taken together, we dem-onstrate that the 3Cs technology enables the rapid and cloning-free generation of high-quality singleand pooled CRISPR/Cas gRNAs.3Cs uncouples sequence diversity from sequence distributionThe absence of PCR amplification and cloning steps, in combination with the uniform distribution ofthe six GFP-targeting 3Cs-gRNAs, led us to reason that 3Cs may uncouple sequence diversity fromsequence distribution during gRNA library generation. To test this hypothesis, we designed sixdegenerated 3Cs oligonucleotides with increasing numbers of randomized nucleotides to mimicgRNA sequence pools with diversities ranging from 256 to 262,144 individual sequences(Figure 2A). The six pools were applied in parallel 3Cs syntheses on a dU-ssDNA template of pLenti-CRISPRv2 (Figure 2B and Figure 2—figure supplement 1A). Independent of an oligonucleotide’sdiversity, NGS and computational analyses identified all of the individual sequences and uniform dis-tributions with area under the curve values between 0.6 and 0.73 (Figure 2—figure supplement 1Band Supplementary file 2). Despite the uniform distribution, we observed a prominent cytosine (C)bias in the randomized libraries, with C contents of above 40% within the top 10% of the most abun-dant gRNAs (Figure 2C). We reasoned that the C bias is probably due to incomplete phosphorami-dite mixing during oligonucleotide synthesis and should therefore be absent from gRNA librariescontaining nonrandom gRNA sequences (Ellington and Pollard, 2009). To test this hypothesis, wedesigned and generated a nonrandom 3Cs-gRNA library targeting all 105 human DUBs, each withthree gRNAs (DUB library). NGS and nucleotide content analysis confirmed our hypothesis andrevealed the absence of C bias from the nonrandom DUB library (Figure 2C andSupplementary file 3). To correct the randomized libraries for the C bias, we determined the indi-vidual nucleotide frequency at every randomized position and used these frequencies to normalizethe original read counts, leading to improved AUC values and sequence distributions (Figure 2Dand Supplementary file 2) and further confirming the uncoupling of sequence diversity and distribu-tion in 3Cs reactions. Taken together, these findings lead us to conclude that 3Cs is a robust tech-nology that uncouples sequence distribution from sequence diversity and, therefore, is a powerfulalternative technology to conventional gRNA cloning methods for generating gRNA libraries.3Cs-gRNA libraries are of high fidelity: the proliferative essentiality ofhuman DUBsNext, we investigated the performance of 3Cs-gRNA reagents in cellular screenings. To do so, wegenerated infectious lentiviral particles of the 3Cs-gRNA DUB library and applied them to a prolifer-ation screen in non-transformed hTERT–RPE1 cells in biological duplicates (multiplicity ofinfection (MOI) 0.2, coverage 1,000). Two days after lentiviral transduction, cells were either col-lected (day 0, reference time point) or selected by puromycin and kept in culture for 11 days (day11) or 21 days (day 21) in cycling conditions representing at least a 1,000-fold library coverage(Figure 3A). Cells collected at day 0, 11, or 21 were subjected to genomic DNA extraction andamplicon-based NGS library preparation, as has been reported previously (Doench et al., 2016;Koike-Yusa et al., 2014). We performed single-read sequencing on an Illumina NextSeq500 with anaveraged read count per gRNA of above 35,000 across all biological samples and replicates(Supplementary file 4). As in previously reported CRISPR analysis algorithms, and to enable a com-parison of individual time points, we summed all individual gRNA read counts per gene and normal-ized each gene read count per sample to the total number of read counts within that sample(Supplementary file 4) (Li et al., 2014;Spahn et al., 2017). In line with reports of the high experi-mental reproducibility of CRISPR/Cas screenings (Evers et al., 2016;Morgens et al., 2016), wedetermined R2values of 0.95, 0.88, and 0.90 for time points day 0, 11, and 21, respectively (Fig-ure 3—figure supplement 3A–C). Spearman correlation and Shapiro–Wilk confidence tests revealedcorrelations of above 0.88 and p-values of below 0.001, respectively, demonstrating a high experi-mental confidence in the level of gRNA representation (Figure 3B and Figure 3—figure supple-ment 3A–C) (Shapiro and Wilk, 1965). To analyze the reproducibility of our screen in terms of thelevel of gene phenotypes, we applied MAGeCK and PinAPL-Py, two established algorithms for theWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 5 of 31Tools and resources Cell Biology Genetics and Genomics QDPH VHTXHQFH¶¶    GLYHUVLW wt gacgaaacacctagggataacagggtaatgttttaga |||||||| **** |||||||| 4N CTGCTTTGctcggagtNNNNagtagtgacCAAAATCT 256 5N CTGCTTTGctcggagtNNNNNgtagtgacCAAAATCT 1,024 6N CTGCTTTGctcggagNNNNNNgtagtgacCAAAATCT 4,096 7N CTGCTTTGctcggagNNNNNNNtagtgacCAAAATCT 16,384 8N CTGCTTTGctcggaNNNNNNNNtagtgacCAAAATCT 65,536 9N CTGCTTTGctcggaNNNNNNNNNagtgacCAAAATCT 262,144A Bkb1210.524dsDNAssDNA3Cs-dsDNA+++ + + + + +4N 5N 6N 7N 8N 9NC0.0 0.2 0.4 0.6 0.8 1.00.00.20.40.60.81.04N, 0.575N, 0.586N, 0.617N, 0.708N, 0.719N, 0.70gRNAs ranked by abundancecumulative fraction ofNGS reads representedD5N6N4N7N8N9NDUBsDUBs, 0.600.0 0.2 0.4 0.6 0.8 1.0020406080100gRNAs ranked by abundanceC content (%)Figure 2. 3Cs is a robust technology that uncouples sequence diversity from sequence distribution. (A) To determine the sequence distribution of 3Cs-gRNA libraries with increasing gRNA diversity, an increasing number of randomized nucleotides (orange) were incorporated into 3Cs oligonucleotidesto mimic gRNA diversities ranging from 256 to 262,144 sequences (4–9N libraries). A range of four to nine randomized nucleotides (orange) wereintroduced into an NHT gRNA sequence. Randomization of the central nucleotides ensures the replacement of the template I-SceI restriction sitein order to prevent the digestion of correctly synthesized 3Cs synthesis products. (B) The 3Cs synthesis products of the combination of randomizedprimers and pLentiGuide were resolved by gel electrophoresis. (C) The Scatter plot displays ranked gRNA abundances per library against the gRNAcytosine content (C). The gRNA libraries that are shown are derived from (A) and (B) and the library with gRNAs targeted against DUBs (DUBs library).All libraries were processed by I-SceI-dependent removal of template plasmid remnants and subjected to NGS and computational analysis.Importantly, all gRNA libraries were complete, irrespective of their individual gRNA diversity. However, the partially randomized gRNA librariesdisplayed a strong C bias within the most abundant gRNA sequences. In fact, the top 5% of most abundant gRNAs had a C content of above 60%. TheDUBs library did not show this C bias, strongly suggesting incomplete phosphoramidite mixing during oligonucleotide synthesis as the main cause ofthe C bias. (D) Lorenz curves displaying the cumulative fraction of represented NGS reads versus the gRNAs ranked by abundance of each partiallyrandomized (4–9N) and nonrandomized (DUB) library revealed a uniform distribution of gRNA sequences. Area under the curve values (AUC, numbernext to library name) confirm the uniform gRNA distribution of these libraries and demonstrate that 3Cs uncouples sequence diversity from sequencedistribution.DOI: https://doi.org/10.7554/eLife.42549.004Figure 2 continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 6 of 31Tools and resources Cell Biology Genetics and Genomics analysis of CRISPR/Cas screens, to raw gRNA read counts of both replicates and calculated aggre-gated positive and negative proliferation phenotypes by means of log2-fold changes with associatedp-values (Figure 3C–E and Figure 3—figure supplement 3D–H and Supplementary file 5–6) (Li et al., 2014;Spahn et al., 2017). Consistent over both time points, cells depleted of USP28 orBRCC3 proliferated more rapidly than cells harboring non-human target sequences (NHTs), identify-ing both as negative regulators of hTERT–RPE1 proliferation (Figure 3C–D and Figure 3F andSupplementary file 4–5). By contrast, cells that were depleted of PSMD14, USP7 or COPS6 prolifer-ated less rapidly than cells harboring non-human target sequences (NHTs), identifying them as posi-tive regulators of hTERT–RPE1 proliferation (Figure 3C–D and Figure 3F and Supplementary file4–5).CRISPR/Cas drop out screens are performed with varying experimental durations, ranging from 5to 15 days (Joung et al., 2017;Potting et al., 2018). However, recent work demonstrates thatCRISPR/Cas induces a G1phase arrest in p53 proficient hTERT–RPE1 cells that impacts hit calling(Haapaniemi et al., 2018), suggesting that later screening time points are beneficial for hit calling.Indeed, when comparing normalized gene ranks, we observed a trend of increased phenotype reso-lution among negative gene ranks over time, although this effect was largely absent from positivegene ranks (Figure 3—figure supplement 1I and Figure 3—figure supplement 2 and 3). This effecthas been reported previously and can potentially be explained by the disproportional assay windowof positive and negative cell proliferation phenotypes, leading to a higher phenotypic resolutionamong negative proliferative effects (Shalem et al., 2014;Wang et al., 2014).Stable and robust proliferative phenotypes are time-independent, so the phenotype resolutionenhances over time (Figure 3—figure supplement 3I). However, multiple mechanisms and cellularbackgrounds can influence phenotype strength, timing and orientation. To identify time-dependentphenotypes, we analyzed the MAGeCK-derived log2-fold changes with associated p-values for timepoints day 11 and day 21, and identified genes whose deletion phenotype significantly changedbetween day 11 and day 21 (Figure 3E–F). Depletion of USP28 and USP46 induced the strongestpositive change, whereas deletion of USP22, USP48 or TNFAIP3 induced the most significant nega-tive change, in phenotype between day 11 and day 21, suggesting a time-dependent absence ofcompensatory mechanisms to accommodate an early loss-of-function phenotype (Figure 3E–F andFigure 3—figure supplement 3I). In order to validate our findings, we chose two positive and nega-tive proliferation-inducing DUBs, generated lentiviral supernatant to deliver shRNA sequences tar-geting the selected DUBs, and transduced hTERT–RPE1 cells. Over the course of the 2 weeks aftertransduction, we measured cell numbers by AlamarBlue staining. When compared to negative-(Luciferase) and positive-control (Plk1) shRNA sequences, depletion of USP28 and BRCC3 induced arapid positive proliferation effect (Figure 3G). By contrast, depletion of USP7 and COPS6 inducedan instant and strong negative proliferation effect (Figure 3G), validating the gRNA-mediatedknockout phenotypes. Collectively, our analysis demonstrates the quality and fidelity of 3Cs reagentsin functional genomics applications.3Cs is versatile and generates arrayed and pooled 3Cs-shRNA reagentsOwing to its versatility,CRISPR/Cas technology has become the method of choice for gene perturba-tion experiments, yet classical short hairpin RNAs (shRNA, RNAi) are still widely used. However,shRNA oligonucleotides contain complementary sequences that form stable secondary structuresthat render the generation of shRNA reagents inefficient (McIntyre and Fanning, 2006). A crucialstep in our improved Kunkel mutagenesis technology is the denaturation of the gRNA-encoding oli-gonucleotides and their subsequent annealing to template ssDNA (see ’Materials and methods’ sec-tion). Owing to this denaturation and annealing step, we anticipated that improved Kunkelmutagenesis would circumvent the secondary structures of shRNA oligonucleotide and enable thegeneration of shRNA reagents. We chose pLKO.1 (Stewart et al., 2003), one of the most widelyFigure 2 continuedThe following figure supplement is available for figure 2:Figure supplement 1. Quality control and gRNA distributions of the randomized libraries.DOI: https://doi.org/10.7554/eLife.42549.005Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 7 of 31Tools and resources Cell Biology Genetics and Genomics A3Cs DUB gRNA libraryhTERT-RPE1MOI 0.2Cov 1,000day 11harvest cells,gDNA extractionday 0harvest cells,gDNA extractionday 21harvest cells,gDNA extraction+ Puro + Puro2 daysFigure 3, Wegner et al.enriched DUBsUSP17L7USP26USP17L12USP17L26USP15USP17L25USP9XUSP34USP14USP4BRCC3USP9Y0USP46USP28day11day21day11/21depleted DUBsGday 0day 11day 21-20-15-10-50sgRNA representation(log2 fraction of reads)CDEBFUSP17L4USP17L19USP17L5USP22USP48BAP1USP19USP26TNFAIP3OTUD5USP17L27USP10USP17L22USP17L28COPS6USP17L18PSMD14OTUD4USP7USP1USP17L29USP3900day11day21day11/21USP17L280log2 fold change (d0/d11) log2 fold change (d11/21)p-value (log10)day 11 day 11/21-2 -1 0 1 2123-4 -2 0 2 4log2 fold change (d0/d21)1234p-value (log10)positivenegativeday 21USP28BRCC3PSMD14NHTsNHTsUSP9Y-2 -1 0 1 21234p-value (log10)positivenegativeUSP28BRCC3USP7NHTsNHTsUSP9YNHTsNHTspositivenegativeUSP28USP46USP22USP48PSMD14COPS6 COPS6USP7TNFAIP30 5 10 1505101520 LuciferasePlk1USP28BRCC3USP7COPS6Time (days)Fold change in cell no.Figure 3. 3Cs reagents are of high fidelity — the essentiality of human DUBs for cell fitness. (A) Schematic of the performed CRISPR screen.Highlighted are the experimental conditions under which the screen was performed (MOI of 0.2, library coverage of 1,000). In brief, hTERT–RPE1 cellswere transduced with lentivirus for 48 hr in duplicates, after which the cells of one duplicate were harvested (day 0) to extrapolate the baseline gRNAdistribution. Simultaneously, cells of the second duplicate were subject to puromycin selection for 11 days, after which time all cells were harvested (day11), counted and plated back in low density to the original library representation of 1,000-fold coverage. Plated cells remained in cycling conditionsuntil day 21, when all cells were collected (day 21). After harvesting the cells, their genomic DNA was extracted and processed for gRNA NGS. (B)Graph showing the distribution of individual sgRNAs. Means ±standard deviation are highlighted. (C–E) Volcano plots visualizing log2-fold changes ofgene phenotypes and their associated p-values. Data are derived from MAGeCK analyses, corresponding to day 11 (C), day 21 (D) and day11/21 (E).The dashed red line shows p=0.05 with points above the line having p 0.05 and points below the line having p 0.05. Data points with p 0.05 aredisplayed as translucent symbols. Genes of interest are highlighted. (F) Venn diagram of significantly enriched (blue) or depleted (orange) DUBs. Thetime point overlap visualizes DUB genes with time independent (overlap of three) and time dependent (overlap of two) proliferation phenotypes. (G)Fold increase in cell number after shRNA-mediated depletion of target genes. Data are means of duplicates.DOI: https://doi.org/10.7554/eLife.42549.006The following figure supplements are available for figure 3:Figure supplement 1 The 3Cs DUB gRNA screens are highly reproducible.DOI: https://doi.org/10.7554/eLife.42549.007Figure 3 continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 8 of 31Tools and resources Cell Biology Genetics and Genomics used lentiviral and shRNA-expressing plasmids to generate circular dU-ssDNA. Similar to circularssDNA of CRISPR/Cas plasmids, the circular ssDNA of pLKO.1 migrated as a single band in agarosegel electrophoresis (Figure 3—figure supplement 2A). Next, we designed a GFP-targeting 3CsshRNA (3Cs-shRNA) primer consisting of 50and 303Cs homology and two complementary GFP–shRNA sequences separated by a six-nucleotide hairpin sequence (Figure 3—figure supplement2B). We performed two parallel 3Cs reactions using 60 ng and 120 ng of ssDNA, and both reactionsyielded the characteristic 3Cs-DNA band pattern with no major difference in bacterial transformationefficiency between the two tested scales (Figure 3—figure supplement 2C).To demonstrate the successful integration of the GFP–shRNA sequence into pLKO.1, we ampli-fied single bacterial clones carrying 3Cs-DNA of shRNA reactions and analyzed their plasmid DNAby SANGER sequencing. This resulted in the expected GFP–shRNA sequence and the absence ofadjacent nucleotide changes (Figure 3—figure supplement 2D), from which we concluded that 3Csis a versatile technology that generates high-quality gRNA and shRNA reagents. To demonstrate3Cs-shRNA fidelity, we generated infectious lentiviral particles of the GFP-targeting 3Cs-shRNA andtransduced GFP-positive hTERT–RPE1 cells. Strikingly, 96 hr after lentiviral transduction, weobserved a reduction in GFP-fluorescence, confirming the functionality of the 3Cs-shRNAs in cells(Figure 3—figure supplement 2E). Moreover, we investigated the performance of improved Kunkelmutagenesis in generating 3Cs-shRNA libraries. On the basis of the principles described above, wedesigned a 3Cs-shRNA library targeting all human ubiquitin-conjugating E2 enzymes (E2s), each withtwo shRNAs (Supplementary file 7). To generate the library, individually synthesized oligonucleoti-des were pooled in equimolar ratios and applied to a pooled 3Cs reaction. The resulting productswere resolved by gel electrophoresis (Figure 3—figure supplement 3A). Like the I-SceI-mediateddepletion of wildtype remnants from CRISPR/Cas 3Cs-gRNA constructs, a Bsu36I restriction enzymeclean-up step removed pLKO.1 wildtype remnants, and SANGER sequencing of the final E2 3Cs-shRNA library (E2.2) confirmed a randomization of forward- and reverse-complement shRNAsequences (Figure 3—figure supplement 3B–C). To determine the E2 3Cs-shRNA distribution moreaccurately, we performed NGS sequencing with an average shRNA read count of 8,300 and deter-mined a wildtype remnant level of 0.04%, a CV of 37.9% and an AUC of 0.68, demonstrating analmost uniform distribution (Figure 3—figure supplement 3D–E). To correlate 3Cs-shRNA abun-dance and the distribution of the E2 3Cs-shRNA libraries before and after Bsu36I enzyme digest, wedetermined the ratios of their respective normalized read counts. Importantly, all ratios were closeto one and lined up close to the respective diagonal with a linear regression R2of 0.9687 (Figure 3—figure supplement 5F), demonstrating a high correlation of individual data points and no influenceof the Bsu36I digest on 3Cs-shRNA sequence distribution. In summary, this demonstrates that our3Cs technology can be adapted to generate high-quality shRNA reagents in single and pooledformats.A partially randomized 3Cs gRNA library to target the coding andnoncoding genome simultaneouslyThe 3Cs method does not require the PCR-amplification of gRNA-encoding oligonucleotides, is freeof conventional cloning steps and uncouples sequence diversity from sequence distribution. Thus,we hypothesized that 3Cs gRNA library diversity is mostly limited by the number of distinguishableoligonucleotides within a 3Cs reaction and the subsequent bacterial electroporation efficiencies.Limitations in electroporation efficiencies can be overcome by accumulating the individual efficien-cies of multiple parallel reactions, as routinely performed to amplify phage libraries with diversitiesbeyond 109(Smith and Scott, 1993). The number of distinguishable oligonucleotides is limited bythe capacity of synthetic oligonucleotide synthesis, rendering truly genome-wide gene perturbationlibraries unfeasible.Figure 3 continuedFigure supplement 2. 3Cs facilitates the generation of shRNA libraries.DOI: https://doi.org/10.7554/eLife.42549.008Figure supplement 3. A 3Cs E2 shRNA library.DOI: https://doi.org/10.7554/eLife.42549.009Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 9 of 31Tools and resources Cell Biology Genetics and Genomics Previously identified SpCas9 nucleotide preferences included a preference for 30pyridine baseswhereas thymidine nucleotides are disfavored (Doench et al., 2014;Doench et al., 2016). In anexploratory effort, we translated SpCas9 gRNA nucleotide preferences into a degenerated oligonu-cleotide sequence (truly genome-wide, TGW) of 20 nucleotides, representing a theoretical diversityof 7.3 1010(Figure 4A) and maximally targeting 1.65 107sites in the human coding and noncod-ing genome (Figure 4A and Supplementary file 8). As mentioned above, randomized positions inDNA oligonucleotides can contain a strong single nucleotide cytosine bias, we therefore used hand-mixed phosphoramidite pools to generate this oligonucleotide pool. In eight parallel large-scale 3Csreactions (each involving 20 mg ssDNA and 600 ng oligonucleotide), we applied this oligonucleotideto dU-ssDNA of pLentiGuide and pLentiCRISPRv2 and resolved the 3Cs products by gel electropho-resis (Figure 4B). Importantly, we note that the amplification of this degenerated library is limited bythe number of transformed bacteria. As complete generation of this reagent is currently unfeasible,we limited our efforts to eight parallel electroporation reactions and achieved a cumulative transfor-mation efficiency of 1.2 1010, accounting for ~16% of TGW sequences, assuming a stringent uni-form sequence distribution. In order to approximate the gRNA distribution, we generated 14.4million NGS reads and found 94.19% to be unique (Figure 4C and Supplementary file 9). We wenton to extract the nucleotide frequencies for each gRNA position from TGW NGS reads, translatedthem to IUPAC nomenclature, and identified the identical degenerated sequence that we initiallyapplied in the form of the degenerated oligonucleotide pool (Figure 4D). Furthermore, the distribu-tion of TGW read counts had a CV of 0.26% and an AUC of 0.52, suggesting a nearly uniform distri-bution of sequenced and represented gRNA sequences (Figure 4E–F) (Makowski and Soares,2003).The 3Cs technology enables the generation of gRNA libraries with sequence diversities exceed-ing those that can be captured by coverage-based screenings. Being aware of the coverage limita-tions, we explored a TGW library screen in the context of a strong positive selection pressure. InhTERT–RPE1 cells, doxorubicin induces a robust, irreversible and dose-dependent reduction of cellviability within 4 days (Figure 5—figure supplement 1A). We therefore generated 5.5 108infec-tious lentiviral particles of the TGW library and applied them to screen for resistance to doxorubicin(Figure 5A–B). In three biologically independent experiments, we transduced a total of 5.5 108hTERT–RPE1 cells with a MOI of 1, added 1 mM doxorubicin 7 d post transduction and replaced themedia every 7 d for 21 consecutive days. Cells that survived the treatment were harvested and theirgenomic DNA was extracted for NGS (Figure 5B). Although the experimental reproducibility waslow (0.004%), we identified an experimental overlap of 4,232 gRNAs, with associated Spearmanranking and Pearson correlations of above 0.75 (Figure 5C and Figure 5—figure supplement 1B).To validate these sequences, we designed and generated a new 3Cs-gRNA validation library consist-ing of the identified 4,232 gRNAs and repeated the doxorubicin resistance screen with establishedexperimental parameters (coverage of 1,000 and MOI of 0.2) (Figure 5—figure supplement 2A–B).As a result, we reidentified 2,716 gRNAs of which 795 were more than two-fold enriched after 21 dof doxorubicin treatment when compared to the untreated control (Figure 5D andSupplementary file 11). In order to map the 795 gRNA sequences to a location within the humangenome, we applied Cas-OFFinder and used the Ensembl, ENCODE, Roadmap Epigenomics andBlueprint databases for sequence annotation (Bernstein et al., 2010;Dunham et al., 2012;Ferna´ndez et al., 2016;Zerbino et al., 2018). We identified seven gRNAs to target five genes(PDE8B, AVPR2, CYSLTR2, IL3RA, and POLE2), of which PDE8B and AVPR2 were targeted by twogRNAs, and a single gRNA sequence matched a noncoding location within chromosome 8(chr8:93022800) (Figure 5E and Figure 5G and Supplementary file 12–13). The coding hits that weidentified included CysLTR2, a Leukotriene C4 G-protein-coupled eicosanoid receptor that wasrecently reported to induce doxorubicin resistance by abolishing the accumulation of reactive oxy-gen species (Dvash et al., 2015). To validate CysLTR2 as a doxorubicin-resistance inducing hit, wechemically inhibited CysLTR2 with increasing concentrations of Bay-CysLT2 or Bay-u9773 in the pres-ence of doxorubicin and quantified cell viability by AlamarBlue staining. Importantly, both drugsreverted the doxorubicin-induced toxicity in a dose-dependent manner (Figure 5F), suggesting thatthe loss of CysLTR2 causes doxorubicin resistance.To account for sequence differences between rRNAs from hTERT–RPE1 cells and the referencegenome (GRCh38.86), as well as SpCas9 off-target activity (Cho et al., 2014;Hsu et al., 2013;Pattanayak et al., 2013), we extended our computational analysis by allowing up to twoWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 10 of 31Tools and resources Cell Biology Genetics and Genomics mismatches during Cas-OFFinder-based target sequence identification. As expected, the number ofgRNAs that could be mapped to the reference genome increased to 192 and 222 for coding andnoncoding target sites, respectively, accounting for 50.3% of the 795 gRNAs (Figure 5E andSupplementary file 12). Interestingly, when mismatches are allowed, we identified three gRNAsthat targeted two different coding positions within the AKAP6 gene (chr14:32671632,chr14:32784395), as well as five different gRNAs that targeted the exact same coding position withinCABFEgRNAs (%)10010.01.001.0001.000011.000001gRNA abundance (reads, log10)10 100 1,000 10,0001.10123-10.26%gRNA pair copy number(median normalized, log10)ACGTTGW gRNA Position050%1234567891011121314151617181920N N N N N N N N N N N ND DH H H V V RssDNA3Cs-dsDNAkb1210.524++++GuideCRISPRv2pLentiTGW gRNA1DPH6HTXHQFH¶¶   LYHUVLW TGW NNDNNNNNHNNNNHDHNVVR 7.3x1010 0.0 0.2 0.4 0.6 0.8 1.00.00.20.40.60.81.0gRNAs ranked by abundancecumulative fraction ofNGS reads representedAUC, 0.52Figure 4. A truly genome-wide (TGW) CRISPR/Cas 3Cs-gRNA library to interrogate the coding and noncoding genome. (A) Previously reported SpCas9nucleotide preferences were translated into a degenerated oligonucleotide sequence (TGW) representing a total sequence diversity of 7.3 1010(Doench et al., 2014). (B) The TGW oligonucleotide shown in panel (A) was used in a 3Cs reaction on template ssDNA derived from pLentiGuide andplentiCRISPRv2 plasmids to generate 3Cs-dsDNA, which was analyzed by gel electrophoresis. (C) Scatter plot visualizing TGW library NGS data from14,448,469 total reads. Displayed are the log10values of gRNA abundance (reads) against the log10of the respective percentage of identified TGWgRNAs. 94.2% of all identified gRNAs were found only once (see also Supplementary file 9). (D) High-throughput sequencing data from panel (C) wereused to compute the nucleotide frequency at each gRNA nucleotide position in order to determine the nucleotide profile of the TGW library. Theidentified nucleotide frequencies closely resemble the pattern of the degenerated TGW oligonucleotide from panel (A). Color code representsnucleotide frequency as indicated by the color gradient on the right. (E) Box plot of TGW gRNA distribution with data derived from panel (C). Thecoefficient of variation of 0.26% suggests a uniform distribution of represented sequences. (F) The gRNA distribution of the TGW library as derivedfrom panel (C) plotted as a Lorenz curve. TGW NGS data derived from pane; (C). The area under the curve (AUC) of 0.52 suggests a uniformdistribution of gRNA sequences.DOI: https://doi.org/10.7554/eLife.42549.010Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 11 of 31Tools and resources Cell Biology Genetics and Genomics Apooled lentivirus55 mL with 1E7 titer(5.5E8 different TGW 3Cs-gRNAs) TiterdeterminationHEK293Ttransfectionpooled8 x TGW 3Cs-reactions(each1.5E9 electrop. eff.)lentiviralsupernatantBTGW gRNA libraryhTERT-RPE1MOI 17 daysPuromycin+ Doxorubicin(1µM) 3 weeksDoxorubicinharvest cells,gDNA extractionFigure 5, Wegner et al.120117,19332,9494232(0.004%)Exp. 1Exp. 2Exp. 3C DE.01 .05 .1 .25 .5 .75 1Doxorubicin (uM)0110Bay-CysLT2(uM)0110Bay-u9773(uM)01010Survival(AU)Survival(AU)F0 1,000 2,000 3,0000.010.1110100gRNA validation rankgRNA enrichment (%, Doxo / CTRL) = 2 fold 2 foldGchr14: 49663308 (POLE2) chr13: 48707205 (CYSLTR2)chr5: 77412152 (PDE8B)chrX: 1352468 (IL3RA)chrX: 153905834 (AVPR2)chr9: 5403147 (PLGRKT)chr5: 79496284 (HOMER1)chr10: 127289202 (DOCK1)chr12: 77705430 (NAV3)chr17: 65181562 (RGS9)3,6% 2,6%0 1 2mismatch:0,45%chr8: 930228003,6%chr17: 65181562(TF binding site) chr15: 93351922(lincRNA: AC091078.1)chr11: 34717155chr3: 134378450(CTCF Binding Site)chr2: 211103345chr1: 29331667(lincRNA: LINC01756) chr10: 26873904chr7: 1349744900 1 2mismatch:213 (95.9%)H182(93.8%)Coding HitsDown in ERCC3mtCatalyric ActivityRegu lat ionUNKNOWN25/8553.7e-17 5.8e-15 8.3e -15 2.2e-1329/151831/179130/1890lincRNAs (11.3%)Open chromatin (6.8%)Processed pseudogene (4.5%)CTCF binding site (3.6%)Predicted promoter (12.6%)Antisense (5.0%)TF binding site (3.1%)Predicted enhancer (0.4%)No annotation available (52.7%)J KSTK3PCSK9DNMBPASAP2RPS6KA2PPP2R3AMECOMHOMER1EHBP1NAV3FOXN3SGMS1WWC1LDLRDSTSLC16A7USTCHST3GSK3BTAB2ADCY9PDE8ANUP98NUP85SEH1LSEC31ASEC24BNUP107SEC23ATRAPPC8SEC13ELMO1CRKDOCK4DOCK1IBiosynthetic process 33/1805UNKNOWNTissue developmentPhosphorus Metabo.FOXO4StimuliCyclic compound33/189029/151829/161832/206130/192921/9171.9e-14 6.8e-14 3.1e-13 1.4e-12 3.9e-12 1.9e-11 2.8e-11Figure 5. TGW-based identification of coding and noncoding sequences that are associated with doxorubicin resistance. (A) Scheme illustrating theworkflow used to generate the pooled lentivirus of the TGW library. The DNA of eight independent TGW 3Cs syntheses was pooled and used totransfect HEK293T cells to produce 5.5 108infectious lentiviral particles. (B) Experimental workflow of the doxorubicin screen in hTERT–RPE1 cells.hTERT–RPE1 cells were transduced with TGW lentivirus with an MOI of 1, selected with puromycin for 7 days, and treated with 1 mM doxorubicin. AfterFigure 5 continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 12 of 31Tools and resources Cell Biology Genetics and Genomics the ASPA2 gene (chr2:9229295) (Figure 5E and Supplementary file 12). Within the noncodinggRNA target sites, we identified four gRNAs that targeted four different positions on chromosomeX (56546543, 57766898, 63133046, 63245878), all of which are in close proximity to the SPIN2Agene (Figure 5G and Supplementary file 13), suggesting a doxorubicin-tolerance-inducing functionin this locus.In order to reveal whether the identified set of coding genes correlated with reported pheno-types or gene ontologies, we performed a molecular signature analysis of the 178 coding targetregions and identified 25 genes that match the UV_RESPONSE_VIA_ERCC3 (downregulated inmutant ERCC3-expressing fibroblasts) group as the most significant hit (p-value of 6.11E-17)(Figure 5H–I and Supplementary file 14). Importantly, doxorubicin-induced interstrand crosslinksare repaired by ERCC3-dependent nucleotide excision repair (NER), and NER-deficient cells havebeen shown to display greater tolerance to adduct-forming anthracycline treatment, connectingthese 25 genes to an increased doxorubicin tolerance (Bret et al., 2013;Spencer et al., 2008;van Brabant et al., 2000). Furthermore, mutations in noncoding sequences have been linked to themisregulation of adjacently located genes by disrupting cis-regulatory elements (Hnisz et al., 2016;Katainen et al., 2015;Weinhold et al., 2014). Therefore, we searched for available biotypesthat are associated with the identified noncoding target regions and were able to identify targetsites matching ‘predicted promoter’ (12.6%), ‘lincRNAs’ (11.3%) as well as ‘processed pseudogenes’(4.5%) and ‘CTCF binding sites’ (3.6%) annotations (Figure 5J and Supplementary file 13). How-ever, no biotype or genomic annotation was available for 52.7% of the noncoding gRNAs, and wetherefore identified the nearest 50and 30located genes and used them to perform a molecular sig-nature analysis (Figure 5K and Supplementary file 15). Among the four most enriched molecularsignatures, we identified genes that are regulated by the transcription factors FOXO4, KLF1, andNFAT, noting that FOXO4 and NFAT downregulation has previously been reported to increasedoxorubicin tolerance (Figure 5K and Supplementary file 15).In summary, we explored the possibility of generating a partially degenerated SpCas9-gRNAlibrary and its application in positive selection screens. Despite the limitations attributed to the gen-eration of such a reagent and its applicability in cellular screens, we identified previously known andFigure 5 continuedthree weeks of continuous doxorubicin treatment, all surviving cells were collected and processed for further analysis. (C) Genomic DNA derived fromthree independent experiments (n= 3), performed according to the scheme illustrated in panel (B), was used to perform NGS and gRNA sequenceidentification. Computational analysis identified an experimental overlap of 4,232 gRNAs (see also Supplementary file 10). (D) A 3Cs library containingthe experimental overlap of 4,232 gRNAs (the validation library) was generated and screened with an experimental coverage of 1,000 and an MOI of 0.1(similar to the workflow shown in panel (B); see also Figure 5—figure supplement 2). NGS of all surviving cells and computational analysis identified795 gRNAs that were enriched more than two-fold (orange) when compared to an untreated control. (E) Pie chart visualizing the distribution of codingtarget regions with respect to SpCas9 off-target rate (0 to 2 mismatches). A total of 192 gRNAs (22.38% of 795 gRNAs) could be mapped to codingregions. Color code represents degree of nucleotide mismatch. (F) Chemical inhibition of cells rescued by CysLTR2 from doxorubicin-mediated toxicity.hTERT–RPE1 cells were treated for 4 d with increasing concentrations of doxorubicin and two chemical inhibitors of CysLTR2 (Bay-CysLT2 and Bay-u9773) before cellular viability was determined by AlamarBlue assays. Averaged values over three biological replicates (n= 3) in arbitrary units(AU) are displayed. (G) Pie chart visualizing the distribution of noncoding target regions with respect to SpCas9 off-target rate (0 to 2 mismatches). Atotal of 222 gRNAs (27.92% of 795 gRNAs) could be mapped to noncoding regions. Color codingshows the degree of nucleotide mismatch. (H)Molecular signature analysis of coding gRNA target sites identifies a set of genes that are downregulated in cells expressing mutant forms of ERCC3 asthe top hit. From among the 178-coding gRNA target site-associated genes, 25 genes are part of the ERCC3 group (which has a total of 855 genes)with high confidence (p=3.7e-17). (I) A list of the 25 ‘down in ERCC3 mutated cells’ genes (light green), as well as their known first- and second-degreeinteracting genes (grey), identifies cytokinesis (DOCK1/4 genes) and vesicle transport (SEC24B/TRAPPC8 genes) gene interactions. Interaction dataadapted from String 10.5. (J) Pie chart visualizing the distribution of noncoding gRNA target site annotations, including their frequency (as percentagesof total noncoding hits). Please note: for 52.7% of all noncoding gRNA target sites, no annotation is available. (K) Molecular signature analysis ofnoncoding gRNA target sites, using adjacently located genes (one for each, 50and 30). 33 genes, out of the 211 genes analyzed, are part of the‘Biosynthetic process’ group (which includes a total of 1,805 genes) with high confidence (p=3.4e-10).DOI: https://doi.org/10.7554/eLife.42549.011The following figure supplements are available for figure 5:Figure supplement 1. Doxorubicin is toxic in hTERT-RPEI cells and TGW replicates correlate.DOI: https://doi.org/10.7554/eLife.42549.012Figure supplement 2. Quality control of the TGW validation library.DOI: https://doi.org/10.7554/eLife.42549.013Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 13 of 31Tools and resources Cell Biology Genetics and Genomics unknown genes that are presumably linked to doxorubicin resistance. In addition, we identified non-coding sequence regions and their neighboring genes for which gene set enrichment analysesrevealed an enrichment for transcription factors that are connected to increased doxorubicintolerance.An optimized truly genome-wide 3Cs gRNA libraryA library’s sequence diversity and distribution directly dictates the experimental scale for positiveand negative selection screens. Therefore, reducing the size of the TGW library to enable coverage-based screens is highly desirable. In line with this, gRNAs that are truncated to 17 nt have beendemonstrated to maintain on-target efficiencies while reducing off-target effects (Fu et al., 2014;Wyvekens et al., 2015). We therefore truncated the degenerated TGW oligonucleotide sequenceto 17 nt(optimized TGA, oTGW), approximating to a a total oligonucleotide diversity of 1.5 109(Figure 6A). Importantly, the oTGW sequence diversity is 50-times smaller than the TGWsequence diversity, while the 1.65 107unique target sequences in the human genome remainidentical (Figure 6A and Supplementary file 8). As for the TGW oligonucleotide, we used hand-mixed phosphoramidite pools to synthesize the oTGW oligonucleotide and performed 3Cs reactionsby combining this nucleotide with a ssDNA dU-template of the three conventionally used lentiviralCRISPR/Cas plasmids: pLentiGuide, pLentiCRISPRv2(Puro) and pLentiCRISPR(GFP-Puro). We thendetermined successful 3Cs reactions by gel electrophoresis (Figure 6B). Subsequent to bacterialamplification, an I-SceI clean-up step was performed before the three oTGW libraries were analyzedby NGS with an average of 28.4 million reads per library (Figure 6C–E and Supplementary file 8).Importantly, extracted gRNA-position nucleotide frequencies were extracted and translated toIUPAC nomenclature, revealing the initial oTGW degenerated oligonucleotide sequence (Figure 6F–H). Furthermore, an average wildtype remnant rate of 0.2% was determined and AUC values were0.54 or below (Figure 6—figure supplement 1), suggesting a uniform distribution of representedgRNA sequences in all three oTGW libraries (Makowski and Soares, 2003). Thus, these oTGWlibraries are the first of their kind, have the potential to elevate functional genomics approaches andwill be made available to the scientific community by the Goethe University Depository (http://www.innovectis.de/INNOVECTIS-Frankfurt/Technologieangebote/Depository).DiscussionIn the present study, we describe the 3Cs technology, an improved Kunkel mutagenesisprotocol that facilitates the one-step and cloning-free generation of high-fidelity CRISPR/Cas andRNAi gene perturbation reagents. 3Cs uncouples sequence diversity from sequence distribution,making it useful for the generation of CRISPR/Cas gRNA libraries of arbitrary sequence diversities.The 3Cs technology has several unique features. First, the bacteriophage-mediated generation ofssDNA makes the technology applicable to all plasmids containing a f1-origin of replication. Second,ssDNA-mutagenic oligonucleotides are annealed to the ssDNA of the template DNA, thereby cir-cumventing the need for two oligonucleotides per gRNA and amplification by PCR, reducing associ-ated costs and sequence bias. In line with this, T7 DNA polymerase, which is used in the 3Csreaction, has an error rate of approximately 15 106, resulting in as little as 0.0015% of mutatedheteroduplex 3Cs product assuming 2 mg ssDNA of a 10 kb plasmid (Kong et al., 1993). Third, thepresence of a gRNA placeholder sequence enables the near-complete removal of wildtype plasmidremnants. Last, we demonstrate 3Cs applicability and performance using the example of lentiviralplasmids. However, we foresee the plasmid range to be expanded to recombinant Adeno-Associ-ated Virus (rAAV) plasmids and adenoviruses, as well as coding sequences for protein mutagenesisthat enable in-cell and in-vivo functional screenings.We demonstrate the fidelity and performance of 3Cs reagents by identifying the proliferativephenotype of human DUBs, validating previously known and uncovering hitherto unknown DUB phe-notypes. We show that depletion of the DUBs USP28 and BRCC3 induces positive proliferation phe-notypes, suggesting that they have tumor suppressive functions. In line with this, USP28 was recentlyidentified as preventing p53 elevation in response to centrosome loss resulting from Plk4 inhibition,thereby preventing growth arrest in response to prolonged mitosis (Meitinger et al., 2016). On theother hand, we identify DUB enzymes whose depletion reduces cell fitness dramatically. Amongthem are COPS6 and USP7, both of which have been implicated in DNA damage response andWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 14 of 31Tools and resources Cell Biology Genetics and Genomics EDkb1210.524ACGToTGW gRNA Position050%1234567891011121314151617N N N N N N N N N NDH H H V V RI-SceIScaII-SceI+ScaI-I-SceI-I-SceI-empty P1 P2pLentiGuidekb1210.524I-SceIEcoR-VI-SceI+EcoR-V-I-SceIEcoR-VI-SceI+EcoR-V-I-SceIEcoR-VI-SceI+EcoR-V-empty P1 P2pLentiCRISPRv2-Puro1kb Plus1kb PlusHGACGToTGW gRNA Position050%1234567891011121314151617N N N N N N N N N NDH H H V V RScaII-SceI+ScaIScaII-SceI+ScaICBAdsDNAssDNA3Cs-dsDNAkb1210.524+++++++++Guide v2-Puro v2-GFP/PuroFkb1210.524I-SceIEcoR-VI-SceI+EcoR-V-I-SceIEcoR-VI-SceI+EcoR-V-I-SceIEcoR-VI-SceI+EcoR-V-empty P1 P2pLentiCRISPRv2-GFP/Puro1kb PlusACGToTGW gRNA Position050%1234567891011121314151617N N N N N N N N N NDH H H V V R1DPH6HTXHQFH¶¶   LYHUVLW oTGW ---NNNNNHNNNNHDHNVVR 1.5x109Figure 6. Optimized TGW (oTGW) libraries for functional interrogations in the coding and noncoding genome. (A) oTGW oligonucleotide sequence,based on reported SpCas9 nucleotide preferences. The truncation of three 50nucleotides results in 17-mer gRNAs with a total oligonucleotide diversityof 1.5 109. (B) oTGW 3Cs-dsDNA was synthesized on a ssDNA-template of pLentiGuide, pLentiCRISPRv2-Puro and pLentiCRISPRv2-GFP/Puro. 3Csproducts are analyzed by gel electrophoresis on a 0.8% TAE/agarose gel. (C–E) Removal of template plasmid remnants with an I-SceI restrictionenzyme digest. oTGW 3Cs-dsDNA was electroporated with efficiencies above 6.31 109and amplified for DNA purification (P1). A subsequent I-SceIFigure 6 continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 15 of 31Tools and resources Cell Biology Genetics and Genomics enhanced p53 stability, leading to a prolonged G1-phase cell cycle arrest (Li et al., 2004). In linewith this, recent work demonstrates a direct connection between CRISPR/Cas gene editing and ap53-dependent DNA damage response that is associated with G1cell cycle arrest(Haapaniemi et al., 2018), suggesting that DNA damage-associated DUBs (e.g. COPS6 or USP7)can be used as p53 alternatives to control for DNA-damage-induced cell cycle arrest and negativehit calling in CRISPR/Cas functional genomic screens.CRISPR/Cas functional genomic screens are widely used to interrogate protein-coding regions,but only few studies have used CRISPR/Cas gene editing to investigate the noncoding genome(Canver et al., 2015;Diao et al., 2016;Korkmaz et al., 2016;Sanjana et al., 2016). Althoughthese studies have the potential to open a new area of functional genomics, their general applicabil-ity is limited to a predefined set of gRNAs and therefore to a small subset of genomic regions. Aunique feature of the 3Cs technology is the uncoupling of gRNA sequence diversity from gRNAsequence distribution, facilitating the generation of partially randomized gRNA libraries. A fully ran-domized library with oligonucleotides of length 20 (20N) resembles the entire space of possiblegRNA sequences and comprises 420= 1.1*1012different sequences. Although an oligonucleotidepool covering this huge sequence space could theoretically be synthesized, there are at least twomajor reasons that render the experimental application unfeasible. The first reason is that the frac-tion of gRNA sequences that have a target site in the genome of interest would be very low.The ~300 million SpCas9 target sites in the human genome would be covered by a fully randomizedlibrary, but would represent only 0.027% of the library. Consequently, the second reason is that theexperimental scale would have to be extremely high to cover all naturally occurring target sites,including all non-human targeting gRNAs that also need to be included, with sufficient coverage.Screening the 20N library with a coverage of 100 and a MOI of 0.5 would require 2.2*1014cells, acell number that is clearly not feasible in current experimental setups.By focusing on SpCas9 nucleotide preferences, we introduce the partially randomized TGWlibrary, which preferentially targets active gRNA sequences in the entire genome, including bothcoding and noncoding regions. The size of the TGW library is dramatically smaller than a fully ran-domized library but still comprises 7.3*1010different gRNA sequences. We sought to explore theexperimental application of such a large library and chose to screen for resistance to the cytotoxicagent doxorubicin in hTERT–RPE1 cells. Although insufficient TGW coverage led to low biologicalreproducibility, we were still able to retrieve gRNA overlap of three experiments. We identified pro-tein-coding genes that have previously been associated with doxorubicin resistancesuch as CysLTR2, whose inhibition by two small chemical compounds reverted the doxorubicin-induced toxicity.Interestingly, about half of the gRNAs for which we were able to identify a matching sequence inthe human genome map to regions in the noncoding genome. Noncoding mutations have beenshown to play pivotal roles in tumorigenesis by disrupting the function of cis-regulatory elements(e.g. promoters, enhancer, or transcription- factor binding sites) and topologically associatingdomains (TADs), thereby directly affecting the transcriptional regulation of adjacently located genes(Katainen et al., 2015;Weinhold et al., 2014). We annotated noncoding regionsthat are associated with hits from our TGW screen by mapping the corresponding gRNA sequencesagainst the human reference genome. By allowing up to two mismatches, we attempted to accountnot only for exact matches but also for mismatched target sites. This approach yielded a number ofFigure 6 continuedrestriction enzyme digest and an electroporation of P1 yielded the final 3Cs libraries containing no detectable template plasmid (P2). An analyticalrestriction enzyme digest with I-SceI and EcoRV removes a 2.5-kb DNA fragment from the template plasmid (empty) and to a minor degree from P1DNA pools. No 2.5-kb fragment could be observed in the final P2 DNA library pools, demonstrating the high purity of the final libraries (see alsoFigure 6—figure supplement 1). (F–H) High-throughput sequencing data derived from panels (C–E) were used to compute the nucleotide frequencyof each gRNA nucleotide position, which are visualized as heat maps. The identified nucleotide frequencies closely resemble the pattern of thedegenerated oTGW oligonucleotide shown in panel (A). Color coding illustrates the nucleotide frequencies (0% in blue to 50% in red).DOI: https://doi.org/10.7554/eLife.42549.014The following figure supplement is available for figure 6:Figure supplement 1. oTGW quality control.DOI: https://doi.org/10.7554/eLife.42549.015Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 16 of 31Tools and resources Cell Biology Genetics and Genomics potential hits that should be interpreted with caution because cut sites might be called incorrectlyand more stringent validation criteria are necessary. Nevertheless, we found a small set of gRNAsthat presumably target predicted promoter sequences as well as CTCF and transcription factor bind-ing sites, although further validation is necessary to gain mechanistic insights into how these sequen-ces are linked to doxorubicin resistance.Furthermore, our computational approach for coding and noncoding hit-calling is sensitive toincorrect calling of cut sites that can lead to false-positive target regions. We therefore suggestthat the actual genome sequence of the cell line of interest is used in order to limit false-positive hitcalling. We used the human reference genome (GRCh38.86), which might differ from the hTERT–RPE1 genome, potentially giving misleading conclusions in terms of hit calling. Another strategy toincrease the rate of true-positive hits is to increase library coverage in the experiment. However, it iscurrently not feasible to screen either the 20N or the TGW library with sufficient coverage, at leastnot in adherent cell lines. To enable coverage-based truly genome-wide screenings, we reduced theTGW library diversity by truncating the library to 17-mers, yielding optimized TGW (oTGW) librariesthat are more suited for high-throughput experiments with suspension cells. We propose that theoTGW CRISPR/Cas gRNA libraries are suitable for broad biological screenings with the highestgenetic target complexity. Such screens can be followed by targeted validation screens using newlysynthesized libraries that are tailored to the initially identified cut sites, which will enable rapid func-tional validation of such complex genetic experiments.Materials and methodsKey resources tableReagent type(species) orresource Designation Source or reference RRID identifiersCell line(human)HEK293T ATCC RRID:CVCL_0063Cell line(human)hTERT-RPE1 ATCC RRID:CVCL_4388Cell line(human)RPE1 Ian CheesemanAntibody Anti-GFP (B-2) Santa CruzBiotechnologyRRID:AB_627695Antibody Anti-alpha Tubulin DSHB RRID:AB_2315509Antibody Goat anti-Mouse IgG(H + L) Secondary AntibodyThermo FisherScientificRRID:AB_228307Antibody Goat anti-Rabbit IgG(H + L) Secondary AntibodyThermo FisherScientificRRID:AB_228341Bacteria (E. coli) K12 CJ236 NEB (E4141)Bacteria (E. coli) 10 beta NEB (C3020K)RecombinantDNA reagentpLentiGuide Addgene (52963)RecombinantDNA reagentpLentiCRISPRv2 Addgene (52961)RecombinantDNA reagentPLKO.1 Addgene (8453)RecombinantDNA reagentpPax2 Addgene (12260)RecombinantDNA reagentpMD2.G Addgene (12259)Commercial kit E.Z.N.A. M13 DNAMini KitOmega Bio-Tek(D69001-01)Commercial kit GeneJET Gelextraction kitThermo Fisher(K0692)Continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 17 of 31Tools and resources Cell Biology Genetics and Genomics ContinuedReagent type(species) orresource Designation Source or reference RRID identifiersCommercial kit Plasmid Maxi Kit Qiagen (12163)Commercial kit PureLink GenomicDNA Mini KitInvitrogen(K1820-01)Chemicalcompound, drugAmpicillin Roth (K029.2)Chemicalcompound, drugChloramphenicol Roth (3886.1)Chemicalcompound, drugKanamycin Roth (T832.3)Chemicalcompound, drugNaCl Roth (31434)Chemicalcompound, drugATP NEB (756)Chemicalcompound, drugDTT Cell SignalingTechnologyEurope (7016)Chemicalcompound, drugdNTP mix Roth (0178.1/2)Chemicalcompound, drugPenicillin-streptomycin Sigma-Aldrich(P4333)Chemicalcompound, drugHygromycin CapricornScientific(HYG-H)Chemicalcompound, drugLipofectamin 2000 ThermoFisher(11668019)Chemicalcompound, drugPolybrene Sigma Aldrich(H9268)Chemicalcompound, drugDoxorubicin Selleckchem(S1208)Chemicalcompound, drugBay-CysLT2 Cayman Chemical(10532)Chemicalcompound, drugBay-u9773 Tocris Bioscience(3138)Other T4 DNA ligase NEB (M0202)Other T7 DNA polymerase (unmodified) NEB (M0274)Other 2 mm electroporation cuvette BTX (45–0125)Other Gene Pulser electroporation system BioRad (164–2076)Other I-SceI NEB (R0694)Other DMEM Thermo Fisher(41965–039)Other DMEM/F12 Thermo Fisher(11320–074)Other FBS Thermo Fisher(10270)Other M13KO7 helper phage NEB (N0315)Other Polyethylene glycol Roth (263.2)Other SOC outgrowth medium Thermo Fisher(15544034)Other 2YT medium Roth (6676.2)Other T4 polynucleotide kinase NEB (M0201)Continued on next pageWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 18 of 31Tools and resources Cell Biology Genetics and Genomics ContinuedReagent type(species) orresource Designation Source or reference RRID identifiersOther T7 endonuclease NEB (M0302)Other OneTaq DNA polymerase NEB (M0480)Other Next High-Fidelity 2x PCR Master Mix NEB (M0541)Other NextSeq 500 IlluminaSoftware,algorithmbcl2fastq Illumina RRID:SCR_015058Software,algorithmcutadapt v.1.15 Martin, 2011 RRID:SCR_011841Software,algorithmCasOFFinder v2.4 Bae et al., 2014Software,algorithmSnpEff 4.3T Cingolani et al., 2012 RRID:SCR_005191Software,algorithmMAGeCK Li et al., 2014Cloning of 3Cs template plasmidsThe NHT and I-SceI gRNA sequences (see ’DNA oligonucleotides’) were annealed and cloned intopLentiGuide (Addgene 52963) and pLentiCRISPRv2 (Addgene 52961) via BsmBI restriction enzymedigest (NEB, R0580) and subsequent ligation with T4 ligase (NEB, M0202). Correct clones were iden-tified by Sanger sequencing at Microsynth SeqLab, Switzerland, using a U6 primer (see ’DNAoligonucleotides’).3Cs oligonucleotide designAll of the 3Cs oligonucleotides that were used in experiments are listed in ’DNA oligonucleotides’.DNA oligonucleotides were purchased from Sigma-Aldrich and Integrated DNA Technologies (IDT)as single or pooled oligonucleotides, and from Twist Bioscience or CustomArray Inc. as oligonucleo-tide pools. The 3Cs oligonucleotides were designed with two homology regions flanking theintended 20-nt gRNA sequence. The homology regions were at least 15 nt in length (Tmabove 50˚C)and matched the 30end of the U6 promoter region and the 50start of the gRNA scaffold in the tem-plate plasmids. The TGW and the oTGW 3Cs oligonucleotides were designed on the basis of a pat-tern of nucleotide preferences as previously determined (Doench et al., 2014;Doench et al., 2016).The observed nucleotide preferences were translated into a degenerated 17-nt DNA sequence(oTGW, see ’DNA oligonucleotides’). The randomized oligonucleotides for the six libraries ofincreasing diversity each had stretches of an increasing number of fully randomized nucleotides (see’DNA oligonucleotides’). The oligonucleotide with four randomized positions was designed to con-tain the stretch of four consecutive Ns beginning at position 30 of the oligonucleotide. Oligonucleo-tides with increasing randomization were designed by extending the randomized pattern in analternating fashion left and right by one randomized position each. The randomized segments andthe flanking constant regions were designed to replace the I-SceI recognition site in the templateplasmid to enable the clean-up digestion step. In general, every gRNA was designed to avoidthe occurrence of the I-SceI recognition site.Overview of reagents and equipment needed for the synthesis of 3CsEquipmentDesktop microcentrifuge, shaking incubator at 37˚C, 1.5 ml collection tubes, filtered sterile pipettetips, thermoblocks at 90˚C and 50˚C (e.g., Thermo Fisher, 88870004), an ultracentrifuge capable ofspinning 50 ml falcon tubes at 10,000 rpm (Beckman Coulter Avanti J-30 I ultracentifuge and a Beck-man JA-12 fixed angle rotor), falcon tubes (polypropylene, 50 ml (Corning 352070)), a Bio-Rad GenePulser electroporation system (BioRad 164–2076), electroporation cuvettes Plus (2 mm, Model no.620 (BTX)), a gel electrophoresis chamber, and erlenmeyer flasks (glass, 100 ml).Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 19 of 31Tools and resources Cell Biology Genetics and Genomics KCM Transformation5x KCM buffer (0.5M KCl, 0.15M CaCl2, 0.25M MgCl2), Escherichia coli strain K12 CJ236 (NEB,E4141), SOC outgrowth medium (ThermoFisher Scientific, 15544034), LB-agar plates supplementedwith 100 mg/ml ampicillin (Roth, K029.2).Phage amplification and ssDNA purification2YT media (Roth, 6676.2), M13KO7 helper phage (NEB, N0315), ampicillin (Roth, K029.2), chloram-phenicol (Roth, 3886.1), kanamycin (Roth, T832.3), uridine (Sigma-Aldrich, U3750), 20% PEG/NaCl(20% polyethylene glycol (Roth, 0263.2), 2.5 M NaCl (Roth, 31434)), Dulbecco’s phosphate-bufferedsaline (PBS, Sigma, D8662), E.Z.N.A. M13 DNA Mini Kit (Omega Bio-Tek, D69001-01), store purifiedphage in PBS at 4˚C.3Cs-DNA synthesis10x TM buffer (0.1 M MgCl2, 0.5 M Tris-HCl, pH 7.5), 10 mM ATP (NEB, 0756), 100 mM DTT (CellSignaling Technology Europe, 7016), T4 polynucleotide kinase (NEB, M0201), 100 mM dNTP mix(Roth, 0178.1/2), T4 DNA ligase (NEB, M0202), T7 DNA polymerase (unmodified) (NEB, M0274),Thermo Fisher Scientific GeneJET Gel Extraction Kit (Thermo Fisher, K0692), 3M sodium acetate(Sigma-Aldrich, 71196).Electroporation and I-SceI clean-up digest2 mm cuvette (BTX, 45–0125), electrocompetent E. coli (10-beta, NEB, C3020K), SOC outgrowthmedium (Thermo Fisher, 15544034), LB-media (Roth, X964.3) supplemented with 100 mg/ml ampicil-lin, Qiagen Plasmid Maxi Kit (Qiagen, 12163), I-SceI (NEB, R0694), NEB CutSmart buffer (NEB,B7204), 0.5% TAE/agarose gel, Thermo Fisher Scientific GeneJET Gel Extraction Kit.dU-ssDNA template amplificationBacteria (Escherichia coli strain K12 CJ236, NEB, E4141) were transformed with 500 ng of templateplasmid according to the following protocol: DNA was mixed with 2 ml of 5x KCM buffer (0.5M KCl,0.15M CaCl2, 0.25M MgCl2) set to 10 ml with water and chilled on ice for 10 min. An equal volumeof CJ236 bacteria was added to the DNA/KCM mixture, gently mixed, and incubated on ice for 15min. The bacteria–DNA mixture was then incubated at room temperature for 10 min, and subse-quently inoculated into 200 ml of prewarmed SOC media (ThermoFisher Scientific, 15544034). Bacte-ria were incubated at 37˚C and 200 rpm for 1 hr and then selected with ampicillin (100 mg/ml) on LB-agar plates overnight at 37˚C.The next morning, a single colony of transformed CJ236 was picked into 1 ml of 2YT media(Roth, 6676.2) supplemented with M13KO7 helper phage (NEB, N0315) to a final concentration of 1108pfu/ml, chloramphenicol (final concentration 35 mg/ml), and ampicillin (final concentration 100mg/ml) to maintain the host F0episome and the phagemid, respectively. Supplementation of uridine(Sigma-Aldrich, U3750) was set to 2.5 mM. After 2 hr of shaking at 200 rpm and 37˚C, kanamycin wasadded to a final concentration of 25 mg/ml to select for bacteria that have been infected withM13KO7 helper phage. Bacteria were kept at 200 rpm and 37˚C for an additional 6 hr before theculture was transferred to 30 ml of 2YT media supplemented with ampicillin (final concentration 100mg/ml) and kanamycin (final concentration 25 mg/ml). After 20 hr of shaking at 200 rpm and 37˚C,the bacterial culture was centrifuged for 10 min at 10,000 rpm and 4˚C in a Beckman JA-12 fixedangle rotor. To precipitate phage particles, the supernatant was transferred to 6 ml (1/5 of culturevolume) PEG/NaCl (20% polyethylene glycol 8,000, 2.5 M NaCl), incubated for 1 hr at room temper-ature and subsequently centrifuged for 10 min at 10,000 rpm and 4˚C in a Beckman JA-12 fixedangle rotor. The phage pellet was resuspended in 1 ml Dulbecco’s phosphate-buffered saline (PBS,Sigma, D8662) and centrifuged at 13,000 rpm for 5 min, before the phage-containing supernatantwas stored at 4˚C. Circular ssDNA was purified from the resuspended phages with the E.Z.N.A. M13DNA Mini Kit (Omega Bio-Tek, D69001-01) according to the manufacturer’s protocol, and purifiedssDNA was stored at 4˚C.3Cs-DNA synthesisThe oligonucleotides that were used for 3Cs reactions and the suppliers are listed separately (see’DNA oligonucleotides’). 3Cs oligonucleotides for specific pools were mixed in equimolar ratios. 600Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 20 of 31Tools and resources Cell Biology Genetics and Genomics ng of pooled oligonucleotides were phosphorylated by mixing them with 2 ml 10x TM buffer (0.1 MMgCl2, 0.5 M Tris-HCl, pH 7.5), 2 ml 10 mM ATP (NEB, 0756), 1 ml 100 mM DTT (Cell Signaling Tech-nology Europe, 7016), 20 units of T4 polynucleotide kinase (NEB, M0201) and water to a total vol-ume of 20 ml. The mixture was incubated for 1 hr at 37˚C.Phosphorylated oligonucleotides were annealed to the circular dU-ssDNA template by adding 20ml of phosphorylation product to 25 ml 10x TM buffer, 20 mg of dU-ssDNA template, and water to atotal volume of 250 ml. The mixture was denatured for 3 min at 90˚C, annealed for 5 min at 50˚C,and cooled down for 5 min at room temperature.3Cs-DNA was generated by adding 10 ml of 10 mM ATP, 10 ml of 100 mM dNTP mix (Roth,0178.1/2), 15 ml of 100 mM DTT, 2000 ligation units of T4 DNA ligase (NEB, M0202), and 30 units ofT7 DNA polymerase (NEB, M0274) to the annealed oligonucleotide–ssDNA mixture. The 3Cs synthe-sis mix was incubated for 12 hr (overnight) at room temperature. The 3Cs synthesis product wasthen affinity purified and desalted using a Thermo Fisher Scientific GeneJET Gel Extraction Kit(Thermo Fisher, K0692) according to the following protocol: 600 ml of binding buffer and 5 ml 3Msodium acetate (Sigma-Aldrich, 71196) were added to the synthesis product, mixed and applied totwo purification columns, which were centrifuged for 3 min at 460 g (2,500 rpm in a Sigma-Aldrich1–14 table top centrifuge). The flow-through was applied a second time to the same purification col-umn to maximize yield and centrifuged for 3 min at 460 g. DNA was eluted in 40 ml warm water. The3Cs reaction product was analyzed by gel electrophoresis alongside the dU-ssDNA template on a0.8% TAE/agarose gel (100 V, 30 min). 3Cs-shRNA libraries were synthesized according to the proto-col described above with the following modifications: in two setups, either 60 ng or 120 ng of circu-lar template dU-ssDNA of pLKO.1 (Addgene: 1864) was used.Electroporations and I-SceI clean-up digestTo generate pool 1 (P1) of a library, 3Cs-DNA constructs were electroporated with a cold 2 mmcuvette (BTX, 45–0125) into electrocompetent E. coli (10-beta, NEB, C3020K) using a Bio-Rad GenePulser with the following settings: resistance 200 Ohm, capacity 25 F, voltage 2.5 kV. 2 mg of DNAwas mixed with 400 ml of freshly thawed cells. Electroporated cells were rescued in 25 ml of pre-warmed SOC media and incubated for 30 min at 37˚C and 200 rpm.After 30 min of incubation, a dilution series was performed to determine the transformation effi-ciency and the number of transformed bacteria. 10 ml of culture was diluted 101to 1012andplated on LB agar plates supplemented with 100 mg/ml ampicillin. The remaining culture was addedto 200 ml LB-media (Roth, X964.3) supplemented with 100 mg/ml ampicillin. Plates were incubatedovernight at 37˚C, the liquid cultures were incubated overnight at 37˚C and 200 rpm. The next day,the electroporation efficiency and the number of transformed bacteria were determined.The plasmid DNA of the overnight liquid cultures was purified using a Qiagen Plasmid Maxi Kit (Qia-gen, 12163) according to the manufacturer’s protocol.To generate the final pool 2 (P2) of a library, 10 mg of purified P1 DNA was digested with 50 unitsI-SceI (NEB, R0694) and 5 ml NEB CutSmart buffer (NEB, B7204) in a reaction volume of 50 ml for 1.5hr at 37˚C. The digestion reaction was subjected to gel electrophoresis on a 0.5% TAE/agarose gel(100 V, 30 min) to separate the undigested 3Cs synthesis product from linearized template plasmid.The band resembling the undigested correct 3Cs synthesis product was purified using a ThermoFisher Scientific GeneJET Gel Extraction Kit according to the manufacturer’s protocol. In a secondstep, the purified 3Cs synthesis product was electroporated according to the electroporation proto-col described above. The final P2 library preparation was purified from liquid culture using a QiagenPlasmid Maxi Kit according to the manufacturer’s protocol and quality controlled with analyticalrestriction enzyme digests. 3Cs-shRNA pools were generated according to the above protocol withthe following modifications. Instead of using I-SceI for the clean-up digestion, we used Bsu36I todigest template plasmid remnants in the first DNA pool (P1). P1 was electroporated using the set-tings described above to yield the final pool (P2). Both 3Cs-shRNA pools were purified from liquidculture using a Qiagen Plasmid Maxi Kit according to the manufacturer’s protocol and were qualitycontrolled with analytical restriction enzyme digests and Sanger sequencing.Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 21 of 31Tools and resources Cell Biology Genetics and Genomics Cell cultureHEK293T cells (ATCC, CRL-3216) were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM,Thermo Fisher Scientific, 41965–039) and hTERT–RPE1 cells (ATCC, CRL-4000 and Ian Cheeseman’s)in DMEM: Nutrient Mixture F-12 (DMEM/F12, Thermo Fisher Scientific, 11320–074), each supple-mented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific, 10270) and 1% penicillin-strep-tomycin (Sigma-Aldrich, P4333) at 37˚C with 5% CO2. In addition, hTERT–RPE1 cells weresupplemented with 0.01 mg/ml hygromycin B (Capricorn Scientific, HYG-H). hTERT–RPE1 cells wereobtained from ATCC/LGC (CRL-4000) and Ian Cheeseman. No method to ensure the state ofauthentication has been applied. Mycoplasma contamination testing was performed immediatelyafter the arrival of the cells and multiple times during the course of the experiments.Cell extracts and antibodiesPreparation of lysates and immunoblot analyses were performed as described previously using Trislysis buffer (50 mM Tris–HCl (pH 7.8), 150 mM NaCl, 1% IGEPAL CA-630) containing 20 mM NaF, 20mM b-glycerophosphate, 0.3 mM Na-vanadate, 20 mg/ml RNase A, 20 mg/ml DNase and 1/300 pro-tease inhibitor cocktail (Sigma-Aldrich, P8340) and phosphatase inhibitor cocktail #2 (Sigma-Aldrich,P5726) (Kaulich et al., 2015). The antibodies used in this study were purchased from the followingsources: mouse anti-GFP (GFP (B-2): sc-9996, 1:2,000, Santa Cruz Biotechnology, Inc.), mouse anti-Tubulin (clone 12G10, 1:1,000, Developmental Studies Hybridoma Bank, University of Iowa). Second-ary antibodies used for western blot analysis were goat anti-mouse (Thermo Scientific, 31430) andgoat anti-rabbit (Thermo Scientific, 31460). The mouse anti-Tubulin hybridoma cell line (clone#12G10) was developed by J. Frankel and E.M. Nelson under the auspices of the NICHD and main-tained by the Developmental Studies Hybridoma Bank. Protein levels were visualized with PierceECL Western Blotting Substrate on a BioRad ChemiDoc MP imaging system and analyzed with Bio-Rad Image Lab software (version 4.1 build 16).Generation and quantification of lentiviral particlesThe day before transfection, HEK293T cells were seeded to 5 105cells/ml. To transfect HEK293Tcells, transfection media containing 1/10 of culture volume Opti-MEM I (Thermo Fisher Scientific,31985–047), 10.5 ml Lipofectamin 2000 (Thermo Fisher Scientific, 11668019), 1.65 mg/ml transfer vec-tor, 1.35 mg/ml pPAX2 (Addgene, 12260) and 0.5/ml mg pMD2.G (Addgene, 12259) was prepared.The mixture was incubated for 30 min at room temperature and added drop-wise to the media. Thenext morning, the transfection medium was replaced with fresh media to remove the transfectionreagent. Lentiviral supernatant was harvested at 24 hr and 48 hr after transfection, pooled andstored at 80˚C.To determine the lentiviral titer, hTERT–RPE1 cells were plated in a 24-well plate with 20,000 cellsper well. The following day, cells were transduced using 8 mg/ml polybrene (Sigma, H9268) and aseries of 0.5, 1, 5, and 10 ml of viral supernatant. After 3 days of incubation at 37˚C, the percentageof fluorescence-positive cells was determined by flow cytometry. The following formula was used tocalculate the viral titer: Virustiter transducing units=mLð Þ ¼ 20:000 target cells%of GFP positive cells100volume of supernatant mLð Þ .Alternatively, lentiviral titers were determined by colony formation titering assay for lentivirus.Flow cytometryAll samples were analyzed on a FACSCanto II flow cytometer (BD Biosciences), and data were proc-essed by FlowJo software (FlowJo, LLC). Gating was carried out on the basis of viable and singlecells that were identified on the basis of their scatter morphology.Lentiviral transductionhTERT–RPE1 cells were seeded at an appropriate density for each experiment with a maximal conflu-ency of 60–70% in DMEM/F12, supplemented with 10% FBS, 0.02 mg/ml hygromycin, and 1% peni-cillin-streptomycin. On the day of transduction, polybrene was added to the media to a finalconcentration of 8 mg/ml. The volume of lentiviral supernatant was calculated on the basis of thediversity of the respective library and of the desired coverage and multiplicity of infection (MOI) ofthe experiment. The number of cells that were transduced at the beginning of an experiment wasWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 22 of 31Tools and resources Cell Biology Genetics and Genomics calculated by multiplying the diversity of the library with the desired coverage and the desired MOI.For example, the parameters for the DUB library screen were set at a coverage of 1.000 and an MOIof 0.2, that is one lentiviral particle per five cells. The total number of cells that were transduced wascalculated as follows: 363 * 1,000 * 5 = 1,815,000. The next morning, the medium was replaced withfresh media and the cells were subjected to antibiotic selection or experimental analysis.Homology arm lengths and 3Cs reaction timesTo test different homology arm lengths, four 3Cs reactions were performed using four different oli-gonucleotides with increasing lengths of homology to the pLentiGuide NHT, according to the 3Cssynthesis protocol described above. The reaction products were analyzed by gel electrophoresis.To monitor the 3Cs synthesis process over time, we annealed the TGW oligonucleotide to thepLentiGuide NHT and generated 3Cs-dsDNA. 2 ml of the reaction was sampled from the reactiontube and transferred to 20˚C at different timepoints from 0 hr to 20 hr. All samples were analyzedtogether by agarose gel electrophoresis. To visualize the kinetics of the 3Cs reactions, 3Cs-dsDNAband intensities were determined and normalized to time point 0 before plotting against the time oftheir harvest using the Bio-Rad Image Lab software (version 4.1 build 16).eGFP gene editing and T7 endonuclease I assayThe efficiency of eGFP gene editing was analyzed by transducing eGFP-expressing hTERT–RPE1cells with 3Cs gRNA constructs based on pLentiCRISPRv2, a subsequent T7 Endonuclease I assay,and immunoblotting. The experiment was performed in triplicates using a control gRNA (NHT), a sin-gle GFP-targeting 3Cs-gRNA (GFP#1) or a pool of six GFP-targeting 3Cs-gRNAs (GFP#1–6). After7 d of incubation at 37˚C without antibiotic selection, cells were trypsinized and the genomic DNAwas purified using a PureLink Genomic DNA Mini Kit (Invitrogen, K1820-01) according to the manu-facturer’s protocol.To assess the genome targeting efficiency of the 3Cs reagents, we analyzed the four cell popula-tions that were transduced with the NHT-gRNA, the GFP#1 gRNA or the GFP#1–6 pool, orthat were not transduced at all. We PCR-amplified the GFP locus with OneTaq DNA polymerase(NEB, M0480) using 1 mg of genomic DNA, 40 mM dNTPs (final concentration), 0.2 mM of each for-ward and reverse amplification primer (see ’DNA oligonucleotides: eGFP T7 forward and eGFP T7reverse’), 10x OneTaq standard buffer, and 2.5 units of OneTaq DNA polymerase. The cycles wereset up as follows: initial denaturation at 94˚C for 3 min, 39 cycles of denaturation at 94˚C for 20 s,annealing at 55˚C for 30 s, strand extension at 68˚C for 2 min, and final strand extension at 68˚C for5 min. The PCR products were analyzed on a 0.8% TAE/agarose gel (100 V, 30 min) and purifiedusing a Thermo Fisher Scientific GeneJET Gel Extraction Kit according to the manufacturer’s proto-col. The T7 endonuclease I digestion was assembled with 6 mg of purified PCR product, 10x NEBu-ffer 2 water to 48 ml, denatured at 95˚C for 5 min, and annealed in two steps from 95–85˚C with 2˚C/second, and from 85–25˚C with 0.1 ˚C/second. To the annealed PCR product, 7 ml of T7 Endo-nuclease I (NEB, M0302) was added and incubated for 15 min at 37˚C. The fragmented PCR prod-ucts were analyzed on a 0.8% TAE/agarose gel (100 V, 30 min) and band intensities weredetermined using the Bio-Rad Image Lab software (version 4.1 build 16).DUB proliferation screenThe DUB proliferation screen was performed in biological duplicates. hTERT–RPE1 cells were trans-duced with lentiviral supernatant with a MOI of 0.2 and a library coverage of 1,000. For each repli-cate and time point, 2.5 million cells were seeded. Cells corresponding to the control time pointwere harvested 2 d post-transduction. All remaining cells were kept in growing and library-diversity-maintaining conditions in the presence of 10 mg/ml puromycin. After 11 d and 21 d, cells were har-vested and their genomic DNA purified and processed for NGS. Validation of DUB screen hit candi-dates was performed in hTERT–RPE1 cells with 3Cs-shRNA-mediated target gene knockdown andthe subsequent assessment of cell proliferation used an AlamarBlue assay (Bio-Rad, BUF012A).Doxorubicin-resistance-screen hTERT–RPE1 cells were treated with increasing concentrations ofdoxorubicin, ranging from 0 to 1,000 nM, for four consecutive days. After 4 d, the treatment wasstopped by changing the medium to doxorubicin-free medium and cells were cultivated for anotherWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 23 of 31Tools and resources Cell Biology Genetics and Genomics 4 d. After a total of 8 d, cell viability was determined and quantified with an AlamarBlue assay (Bio-Rad, BUF012A).To screen for doxorubicin resistance, the TGW library was delivered in triplicates to a total of 5.5108hTERT–RPE1 cells with doxycycline-inducible Cas9 expression via lentiviral transduction at aMOI of 1. Transduced cells were cultured for 7 d in standard medium supplemented with 1 mM doxy-cycline (Sigma-Aldrich) and 10 mg/ml puromycin. At day 7, the medium was changed to selectionmedium containing 1 mM doxorubicin (Selleckchem, S1208). After 3 wk of selection (fresh doxorubi-cin every 4 d), surviving cells were harvested and processed for NGS.NGS of plasmid and genomic DNATo purify genomic DNA, surviving cells were trypsinized and pelleted. Genomic DNA was extractedusing the PureLink Genomic DNA Mini Kit according to the manufacturer’s protocol. For NGS librarypreparation, 100 ng of plasmid or up to 2 mg of genomic DNA per reaction was used in a 50 ml PCRreaction using Next High-Fidelity 2x PCR Master Mix (NEB, M0541) (according to the manufacturer’sprotocol) and 1 ml of 10 mM primers each of forward and reverse primers. Primer sequences arelisted separately (see ’DNA oligonucleotides’). The sequencing primers contained an 8-nt long bar-code sequence, enabling the multiplexing of several samples in a single sequencing run and Illuminaadapter sequences. Thermal cycler parameters were set as follows: initial denaturation at 98˚C for 5min, 19 cycles of denaturation at 98˚C for 30 s, annealing at 55˚C for 30 s, extension at 72˚C for 1min, and final extension at 72˚C for 5 min. PCR products were purified from a 0.5% TAE/agarose gelusing a Thermo Fisher Scientific GeneJet Gel Extraction Kit according to the manufacturer’s proto-col. The purified PCR product was prepared to a final concentration of 2.4 pM in a total volume of2.2 ml and loaded onto a NextSeq 500 sequencer (Illumina), according to the manufacturer’s proto-col. Sequencing was performed with single end reads, 75 cycles and 8 cycles of single index reading.Data processing and analysisAll data obtained from NGS were demultiplexed using the Illumina command line tool bcl2fastq,v2.17. gRNA representation of all libraries was assessed using cutadapt v1.15 (Martin, 2011) andcustom Python scripts. In brief, 30sequencing adapters were trimmed using a prefix of the 30homol-ogy sequence; trimmed reads were further trimmed by keeping only the last 20 nucleotides for alllibraries except the oTGW, for which the last 17 nt were kept. Only reads with no ambiguouslysequenced nucleotides were considered for further analyses. For the TGW and the oTGW, the result-ing sequences were compared to the TGW or oTGW DNA sequence pattern, respectively, usingPython and regular expressions. The reads obtained from sequencing the six randomized librarieswith diversities ranging from 256 to 262,144 gRNAs were processed similarly by comparing thetrimmed reads with the gRNA pattern of the respective library. For the GFP and DUB libraries, thereads were aligned to the respective sequence library. Matching sequences were counted to deter-mine the read count distribution of a sample. The read counts of individual gRNAs for a samplewere normalized by the total number of read counts that could be assigned to the respective library.The screening of samples after treatment of the cells was carried out in the same way. To determinethe dispersion of the read counts, the coefficient of variation was computed by dividing the standarddeviation of the normalized read counts by the mean of the normalized read count x,CV ¼sx. Toassess the uniformity of each library distribution, we generated Lorenz curves of gRNA representa-tion. The Lorenz curves of gRNA representation rank gRNAs by abundance scaled to 1 and show thefraction of total sequencing reads that are represented by the sum of gRNA read counts. The areaunder the curve (AUC) was computed in GraphPad Prism 5.0b for Mac (GraphPad Software, La JollaCalifornia USA, www.graphpad.com) or with a custom Python script using Numpy 1.14.2 (Oli-phant, 2010). Heat maps were generated by accumulating the nucleotide frequency at each positionof the sequenced reads and normalized by the total number of read counts.To correct the read counts of the six randomized libraries with diversities ranging between 256and 262,144 gRNAs for C bias, we determined the nucleotide frequencies for each sequence posi-tion of the trimmed and final reads and normalized the observed frequencies to the expected nucle-otide frequency of 25%. Each read was then scored by summing the normalized frequencies for allreads individually. The observed read count per gRNA was then multiplied with this score, dividedby the sum of all read counts that matched the respective gRNA pattern, and normalized to the sumWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 24 of 31Tools and resources Cell Biology Genetics and Genomics of all corrected and normalized read counts. Lorenz curves were generated on the basis of the cor-rected and normalized read counts.Read count data from the DUB screen were analyzed by summing all individual gRNA read countsper gene and normalizing each gene read count per sample to the total number of read countswithin that sample. Spearman correlation and Shapiro-Wilk confidence tests were performed toassess the reproducibility of the DUB screen replicates. MAGeCK and PinAPL-Py were used to ana-lyze the read counts of both replicates and to calculate aggregated positive and negative prolifera-tion phenotypes by means of log2fold changes with associated p-values.Analysis of gRNA on- and off-target locationsTo determine the on- and off-targets of the 4232 hits from the doxorubicin resistance screen, Cas-OFFinder (v2.4) was applied to search the human genome (GRCh38.86) for gRNA target sites withup to two mismatches (Bae et al., 2014). The genomic positions of each on- and off-target wereannotated with Ensembl genome assembly GRCh38.86, using SnpEff 4.3T (Cingolani et al., 2012)and custom Python scripts. Multiple annotations for a location were collapsed onto a single genetype and the corresponding gene name, if available. Genomic locations associated with an inter-genic region were not considered to be annotated. Additional noncoding, regulatory, and pseudo-gene information was annotated using the Ensembl regulatory and motif features from release 91and the Gencode consensus pseudogenes dataset from release 27 (GRCh38.p10). Additional stan-dard annotation data from Gencode, release 27, were also included. Spearman rank and Pearsoncorrelation were computed with NumPy (1.14.2). To determine the putative effect of gRNA off-tar-gets on previously identified on-target locations, we mapped the gene names that were associatedwith off-targets back to the genes that were associated with on-target hits.Validation of TGW doxorubicin-resistance screening hitsTo validate CysLTR2 as a doxorubicin-resistance-inducing gene, we applied the CysLT2 receptorantagonists Bay-CysLT2 (Cayman Chemical, 10532) and Bay-u9773 (Tocris Bioscience, 3138). In twodifferent triplicates, cells were treated with with increasing concentrations (0 mM, 0.01 mM, 0.05 mM,0.1 mM, 0.25 mM, 0.5 mM, 0.5 mM, 1 mM) of each inhibitor as well as with increasing doxorubicin con-centrations (0 mM, 1 mM, and 10 mM). After 4 d, cell survival was assessed with an AlamarBlue assay.10% AlamarBlue was added to the cultured cells and incubated for 2 hr at 37˚C, and fluorescencewas measured with an excitation wavelength of 560 nm and a fluorescence emission of 590 nm on aBioTek Synergy H1 microplate reader. The given measured fluorescence emissions were averagedover all replications for each experiment.The 4,232 hits that were found in the TGW doxorubicin screen were compiled into an individual3Cs library (validation library). The validation library was generated according to the 3Cs DNA syn-thesis protocol described above. We seeded 1 106hTERT–RPE1 in T175 cell culture flasks inDMEM/F12 and transduced them the next day with lentivirus of the validation library using 8 mg/mlpolybrene with an MOI of 0.1 and an experimental coverage of 1,000. After 3 d, the control cellswere harvested. The screen was conducted with 2.5 mg/ml (final concentration) puromycin selectionand 1 mM doxorubicin treatment. The medium was changed every third day to maintain constantpuromycin and doxorubicin concentrations. After three weeks of selection, surviving cells were har-vested and processed for NGS according to the procedures described above.Molecular signatures of coding and noncoding hitsHits for targets with zero, one and two mismatches were merged and divided in two subsets accord-ing to the annotation, consisting of protein coding hits and noncoding hits, respectively. For thecoding hits gene set, a set of 159 non-redundant genes was created from all hits with target sites inprotein-coding genes. The frequency of each gene was determined. For the remaining hits, the fiveclosest genes upstream and the five closest genes downstream of the target site were determinedusing GRCh.93 Ensembl gene data. The starting position of a gene and the starting position of a tar-get site was taken as measure for proximity. A noncoding-hits gene set of 1,805 non-redundantgenes was created and the frequency of each gene was determined. Overlaps between both genesets and all gene sets in the Molecular Signatures Database (MSigDB) were computed usingthe MiSigDB Web Application to Investigate Gene Sets with a FDR q-value below 0.05Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 25 of 31Tools and resources Cell Biology Genetics and Genomics (Subramanian et al., 2005;Liberzon et al., 2011), Heatmaps with the top four overlapping genesets were created using the resulting Excel tables (Figure 5H,Supplementary files 13–14 and thePython visualization library seaborn 0.9.0 (Waskom et al., 2014).To collect interaction data, wesearched the 25 genes that were associated with ’down in ERCC3 mutated cells’ in the String 10.5database by choosing up to five interactors in the first and second shell (Szklarczyk et al., 2017)(Figure 5I).Data availabilityNGS data are provided as raw read count tables as Supplementary files 1, and 17–23. Please note,raw read count tables associated with TGW and the three oTGW libraries are available from Dryad,https://doi.org/10.5061/dryad.rs432pr. Plasmids encoding oTGW 3Cs-gRNA libraries will be avail-able through the Goethe University Depository (http://www.innovectis.de/INNOVECTIS-Frankfurt/Technologieangebote/Depository).Code availabilityCustom software is publicly available from GitHuB, https://github.com/GEG-IBC2/3Cs (GEG-IBC2, 2019; copy archived at https://github.com/elifesciences-publications/3Cs).DNA oligonucleotidesDNA oligonucleotides were purchased from Sigma-Aldrich and Integrated DNA Technologies (IDT)as single or pooled oligonucleotides, and from Twist Bioscience or CustomArray Inc. as oligonucleo-tide pools. A detailed list of all oligonucleotides can be found as supplementary information(Supplementary file 16).AcknowledgementsWe thank Ian Cheeseman and Kara Lavidge McKinley for providing Cas9-inducible hTERT–RPE1cells. We thank Tony Gutschner and Christian Mu¨nch for critical reading and commenting on themanuscript. We thank Tobias Schmidt for valuable advice and support with NGS. The hybridoma(12G10, alpha-tubulin) developed by the University of Iowa was obtained from the DevelopmentalStudies Hybridoma Bank, created by the NICHD of the NIH and maintained at The University ofIowa, Department of Biology, Iowa City, IA 52242, USA. This work was supported by the HessianMinistry for Science and the Arts (HMWK, LOEWE-CGT, IIIL5-518/17.004), the German ResearchFoundation (DFG; CEF-MC - EXC115/2; ECCPS - EXC147/2) and in part by the LOEWE CenterFrankfurt Cancer Institute (FCI) funded by the Hessen State Ministry for Higher Education, Researchand the Arts (IIIL5-519/03/03.001-(0015)).Additional informationCompeting interestsIvan Dikic: is co-founder, shareholder and CEO of Vivlion GmbH in Gru¨ndung. Also a senior editor ofeLife. Martin Wegner: The Goethe University Frankfurt has filed a patent related to this work onwhich MW is an inventor (WO2017EP84625). Valentina Diehl: The Goethe University Frankfurt hasfiled a patent related to this work on which VD is an inventor (WO2017EP84625). Rahel de Bruyn:The Goethe University Frankfurt has filed a patent related to this work on which RDB is an inventor(WO2017EP84625). Svenja Wiechmann: The Goethe University Frankfurt has filed a patent related tothis work on which SW is an inventor (WO2017EP84625). Andreas Ernst: The Goethe UniversityFrankfurt has filed a patent related to this work on which AE is an inventor (WO2017EP84625). Man-uel Kaulich: The Goethe University Frankfurt has filed a patent related to this work on which MK isan inventor (WO2017EP84625). Also co-founder, shareholder and CSO of Vivlion GmbH in Gru¨n-dung. The other authors declare that no competing interests exist.Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 26 of 31Tools and resources Cell Biology Genetics and Genomics FundingFunder Grant reference number AuthorHessisches Ministerium fu¨rWissenschaft und KunstIIIL5-518/17.004 Manuel KaulichDeutsche Forschungsge-meinschaftEXC115/2 Manuel KaulichHessisches Ministerium fu¨rWissenschaft und KunstIIIL5-519/03/03.001 Manuel KaulichDeutsche Forschungsge-meinschaftEXC147/2 Manuel KaulichThe funders had no role in study design, data collection and interpretation, or thedecision to submit the work for publication.Author contributionsMartin Wegner, Resources, Software, Formal analysis, Validation, Investigation, Visualization, Meth-odology, Writing—original draft, Writing—review and editing; Valentina Diehl, Formal analysis, Vali-dation, Investigation, Visualization, Methodology, Writing—original draft; Verena Bittl, Investigation,Visualization, Writing—original draft; Rahel de Bruyn, Investigation, Visualization, Methodology;Svenja Wiechmann, Investigation, Methodology; Yves Matthess, Investigation, Methodology, Writ-ing—original draft, Writing—review and editing; Marie Hebel, Software, Investigation, Visualization,Methodology; Michael GB Hayes, Sven Heinz, Resources, Investigation; Simone Schaubeck, Valida-tion, Investigation; Christopher Benner, Resources, Software; Anja Bremm, Investigation, Writing—original draft; Ivan Dikic, Supervision, Funding acquisition, Investigation, Writing—original draft,Writing—review and editing; Andreas Ernst, Conceptualization, Methodology, Writing—originaldraft, Writing—review and editing; Manuel Kaulich, Conceptualization, Software, Formal analysis,Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administrationAuthor ORCIDsMartin Wegner http://orcid.org/0000-0001-6403-3926Yves Matthess http://orcid.org/0000-0003-4040-1258Anja Bremm http://orcid.org/0000-0003-1386-0926Ivan Dikic https://orcid.org/0000-0001-8156-9511Manuel Kaulich http://orcid.org/0000-0002-9528-8822Decision letter and Author responseDecision letter https://doi.org/10.7554/eLife.42549.043Author response https://doi.org/10.7554/eLife.42549.044Additional filesSupplementary files.Supplementary file 1. 3Cs-gRNA GFP library - NGS analysis including total read counts.DOI: https://doi.org/10.7554/eLife.42549.016.Supplementary file 2. List of 4–9N 3Cs libraries - NGS analysis including total read counts perlibrary.DOI: https://doi.org/10.7554/eLife.42549.017.Supplementary file 3. 3Cs-gRNA DUB library - NGS analysis including total read counts of DUBlibrary.DOI: https://doi.org/10.7554/eLife.42549.018.Supplementary file 4. 3Cs-gRNA DUB screen - NGS analysis including total read counts andnormalizations.DOI: https://doi.org/10.7554/eLife.42549.019.Supplementary file 5. 3Cs-gRNA DUB screen - NGS analysis using MAGeCK.Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 27 of 31Tools and resources Cell Biology Genetics and Genomics DOI: https://doi.org/10.7554/eLife.42549.020.Supplementary file 6. 3Cs-gRNA DUB screen - NGS analysis using PinAPL-Py.DOI: https://doi.org/10.7554/eLife.42549.021.Supplementary file 7. List of 3Cs-shRNA E2 library sequences and total NGS read counts.DOI: https://doi.org/10.7554/eLife.42549.022.Supplementary file 8. Number of TGW and oTGW target sequences per human chromosome andtotal oTGW NGS read counts per library.DOI: https://doi.org/10.7554/eLife.42549.023.Supplementary file 9. 3Cs TGW library – list of total NGS read counts.DOI: https://doi.org/10.7554/eLife.42549.024.Supplementary file 10. List of 4232 3Cs TGW gRNA sequences derived from the doxorubicinscreen.DOI: https://doi.org/10.7554/eLife.42549.025.Supplementary file 11. 3Cs-gRNA TGW validation screen - NGS analysis including raw read counts,normalizations and ratios.DOI: https://doi.org/10.7554/eLife.42549.026.Supplementary file 12. Annotation list of validated coding TGW hits.DOI: https://doi.org/10.7554/eLife.42549.027.Supplementary file 13. Annotation list of validated noncoding TGW hits.DOI: https://doi.org/10.7554/eLife.42549.028.Supplementary file 14. List of coding hits - molecular signature analysis.DOI: https://doi.org/10.7554/eLife.42549.029.Supplementary file 15. List of noncoding hits - molecular signature analysis.DOI: https://doi.org/10.7554/eLife.42549.030.Supplementary file 16. List of DNA oligonucleotides.DOI: https://doi.org/10.7554/eLife.42549.031.Supplementary file 17. Raw sequencing counts of the six randomized libraries.DOI: https://doi.org/10.7554/eLife.42549.032.Supplementary file 18. DUB library raw sequencing counts.DOI: https://doi.org/10.7554/eLife.42549.033.Supplementary file 19. DUB screen raw sequencing read counts, includes day 0, day 11, and day 21for two replicates.DOI: https://doi.org/10.7554/eLife.42549.034.Supplementary file 20. Raw sequencing counts of the E2-shRNA library.DOI: https://doi.org/10.7554/eLife.42549.035.Supplementary file 21. Raw sequencing reads of the TGS screen triplicates.DOI: https://doi.org/10.7554/eLife.42549.036.Supplementary file 22. Raw sequencing reads of the TGW screen validation library.DOI: https://doi.org/10.7554/eLife.42549.037.Supplementary file 23. Raw sequencing reads of CTRL and treatment samples of the TGW valida-tion screen.DOI: https://doi.org/10.7554/eLife.42549.038.Transparent reporting formDOI: https://doi.org/10.7554/eLife.42549.039Data availabilityAll data generated or analysed during this study are included in the manuscript, supplementary filesor are available through GitHub or Dryad. NGS data and custom software is available as supplemen-tary files and from Dryad and GitHub. Plasmids encoding oTGW 3Cs-gRNA libraries will be madeavailable through the Goethe University Depository (http://innovectis.de/technologien/goethe-depository/).The following dataset was generated:Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 28 of 31Tools and resources Cell Biology Genetics and Genomics Author(s) Year Dataset title Dataset URLDatabase andIdentifierWegner M, Diehl V,Bittl V, de Bruyn R,Wiechmann S,Matthess Y2019 Data from: Circular synthesizedCRISPR/Cas gRNAs for functionalinterrogations in the coding andnoncoding genomehttps://dx.doi.org/10.5061/dryad.rs432prDryad DigitalRepository, 10.5061/dryad.rs432prReferencesArakawa H. 2016. A method to convert mRNA into a gRNA library for CRISPR/Cas9 editing of any organism.Science Advances 2:e1600699. DOI: https://doi.org/10.1126/sciadv.1600699,PMID: 27574704Bae S, Park J, Kim JS. 2014. Cas-OFFinder: a fast and versatile algorithm that searches for potential off-targetsites of Cas9 RNA-guided endonucleases. Bioinformatics 30:1473–1475. DOI: https://doi.org/10.1093/bioinformatics/btu048,PMID: 24463181Bernstein BE, Stamatoyannopoulos JA, Costello JF, Ren B, Milosavljevic A, Meissner A, Kellis M, Marra MA,Beaudet AL, Ecker JR, Farnham PJ, Hirst M, Lander ES, Mikkelsen TS, Thomson JA. 2010. The NIH roadmapepigenomics mapping consortium. Nature Biotechnology 28:1045–1048. DOI: https://doi.org/10.1038/nbt1010-1045,PMID: 20944595Bret C, Klein B, Moreaux J. 2013. Nucleotide excision DNA repair pathway as a therapeutic target in patientswith high-risk diffuse large B cell lymphoma. Cell Cycle 12:1811–1812. DOI: https://doi.org/10.4161/cc.25115,PMID: 23708513Canver MC, Smith EC, Sher F, Pinello L, Sanjana NE, Shalem O, Chen DD, Schupp PG, Vinjamur DS, Garcia SP,Luc S, Kurita R, Nakamura Y, Fujiwara Y, Maeda T, Yuan GC, Zhang F, Orkin SH, Bauer DE. 2015. BCL11Aenhancer dissection by Cas9-mediated in situ saturating mutagenesis. Nature 527:192–197. DOI: https://doi.org/10.1038/nature15521,PMID: 26375006Cho SW, Kim S, Kim Y, Kweon J, Kim HS, Bae S, Kim JS. 2014. Analysis of off-target effects of CRISPR/Cas-derived RNA-guided endonucleases and nickases. Genome Research 24:132–141. DOI: https://doi.org/10.1101/gr.162339.113,PMID: 24253446Cingolani P, Platts A, Wang leL, Coon M, Nguyen T, Wang L, Land SJ, Lu X, Ruden DM. 2012. A program forannotating and predicting the effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome ofDrosophila melanogaster strain w1118; iso-2; iso-3. Fly 6:80–92. DOI: https://doi.org/10.4161/fly.19695,PMID: 22728672Diao Y, Li B, Meng Z, Jung I, Lee AY, Dixon J, Maliskova L, Guan KL, Shen Y, Ren B. 2016. A new class oftemporarily phenotypic enhancers identified by CRISPR/Cas9-mediated genetic screening. Genome Research26:397–405. DOI: https://doi.org/10.1101/gr.197152.115,PMID: 26813977Doench JG, Hartenian E, Graham DB, Tothova Z, Hegde M, Smith I, Sullender M, Ebert BL, Xavier RJ, Root DE.2014. Rational design of highly active sgRNAs for CRISPR-Cas9-mediated gene inactivation. NatureBiotechnology 32:1262–1267. DOI: https://doi.org/10.1038/nbt.3026,PMID: 25184501Doench JG, Fusi N, Sullender M, Hegde M, Vaimberg EW, Donovan KF, Smith I, Tothova Z, Wilen C, Orchard R,Virgin HW, Listgarten J, Root DE. 2016. Optimized sgRNA design to maximize activity and minimize off-targeteffects of CRISPR-Cas9. Nature Biotechnology 34:184–191. DOI: https://doi.org/10.1038/nbt.3437,PMID: 26780180Dunham I, Kundaje A, Aldred SF, Collins PJ, Davis CA, Doyle F, Epstein CB,ENCODE Project Consortium. 2012.An integrated encyclopedia of DNA elements in the human genome. Nature 489:57–74. DOI: https://doi.org/10.1038/nature11247,PMID: 22955616Dvash E, Har-Tal M, Barak S, Meir O, Rubinstein M. 2015. Leukotriene C4 is the major trigger of stress-inducedoxidative DNA damage. Nature Communications 6:10112. DOI: https://doi.org/10.1038/ncomms10112,PMID: 26656251Ellington A, Pollard JD. 2009. Introduction to the synthesis and purification of oligonucleotides. CurrentProtocols in Nucleic Acid Chemistry.DOI: https://doi.org/10.1002/0471142700.nca03cs00Ernst A, Avvakumov G, Tong J, Fan Y, Zhao Y, Alberts P, Persaud A, Walker JR, Neculai AM, Neculai D,Vorobyov A, Garg P, Beatty L, Chan PK, Juang YC, Landry MC, Yeh C, Zeqiraj E, Karamboulas K, Allali-HassaniA, et al. 2013. A strategy for modulation of enzymes in the ubiquitin system. Science 339:590–595.DOI: https://doi.org/10.1126/science.1230161,PMID: 23287719Evers B, Jastrzebski K, Heijmans JP, Grernrum W, Beijersbergen RL, Bernards R. 2016. CRISPR knockoutscreening outperforms shRNA and CRISPRi in identifying essential genes. Nature Biotechnology 34:631–633.DOI: https://doi.org/10.1038/nbt.3536,PMID: 27111720Ferna´ndez JM, de la Torre V, Richardson D, Royo R, Puiggro`s M, Moncunill V, Fragkogianni S, Clarke L, Flicek P,Rico D, Torrents D, Carrillo de Santa Pau E, Valencia A, BLUEPRINT Consortium. 2016. The BLUEPRINT dataanalysis portal. Cell Systems 3:491–495. DOI: https://doi.org/10.1016/j.cels.2016.10.021,PMID: 27863955Fu Y, Sander JD, Reyon D, Cascio VM, Joung JK. 2014. Improving CRISPR-Cas nuclease specificity usingtruncated guide RNAs. Nature Biotechnology 32:279–284. DOI: https://doi.org/10.1038/nbt.2808,PMID: 24463574GEG-IBC2. 2019. 3Cs. 6c815a4. Github. https://github.com/GEG-IBC2/3CsWegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 29 of 31Tools and resources Cell Biology Genetics and Genomics Guschin DY, Waite AJ, Katibah GE, Miller JC, Holmes MC, Rebar EJ. 2010. A rapid and general assay formonitoring endogenous gene modification. Methods in Molecular Biology 649:247–256. DOI: https://doi.org/10.1007/978-1-60761-753-2_15,PMID: 20680839Haapaniemi E, Botla S, Persson J, Schmierer B, Taipale J. 2018. CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nature Medicine 24:927–930. DOI: https://doi.org/10.1038/s41591-018-0049-z,PMID: 29892067Handa P, Varshney U. 1998. Rapid and reliable site directed mutagenesis using Kunkel’s approach. Indian Journalof Biochemistry Biophysics 35:63–66. PMID: 9753863Hnisz D, Weintraub AS, Day DS, Valton AL, Bak RO, Li CH, Goldmann J, Lajoie BR, Fan ZP, Sigova AA, Reddy J,Borges-Rivera D, Lee TI, Jaenisch R, Porteus MH, Dekker J, Young RA. 2016. Activation of proto-oncogenes bydisruption of chromosome neighborhoods. Science 351:1454–1458. DOI: https://doi.org/10.1126/science.aad9024,PMID: 26940867Hsu PD, Scott DA, Weinstein JA, Ran FA, Konermann S, Agarwala V, Li Y, Fine EJ, Wu X, Shalem O, Cradick TJ,Marraffini LA, Bao G, Zhang F. 2013. DNA targeting specificity of RNA-guided Cas9 nucleases. NatureBiotechnology 31:827–832. DOI: https://doi.org/10.1038/nbt.2647,PMID: 23873081Huang R, Fang P, Kay BK. 2012. Improvements to the Kunkel mutagenesis protocol for constructing primary andsecondary phage-display libraries. Methods 58:10–17. DOI: https://doi.org/10.1016/j.ymeth.2012.08.008,PMID: 22959950Joung J, Konermann S, Gootenberg JS, Abudayyeh OO, Platt RJ, Brigham MD, Sanjana NE, Zhang F. 2017.Genome-scale CRISPR-Cas9 knockout and transcriptional activation screening. Nature Protocols 12:828–863.DOI: https://doi.org/10.1038/nprot.2017.016,PMID: 28333914Katainen R, Dave K, Pitka¨nen E, Palin K, Kivioja T, Va¨lima¨ki N, Gylfe AE, Ristolainen H, Ha¨nninen UA, Cajuso T,Kondelin J, Tanskanen T, Mecklin JP, Ja¨rvinen H, Renkonen-Sinisalo L, Lepisto¨A, Kaasinen E, Kilpivaara O,Tuupanen S, Enge M, et al. 2015. CTCF/cohesin-binding sites are frequently mutated in cancer. NatureGenetics 47:818–821. DOI: https://doi.org/10.1038/ng.3335,PMID: 26053496Kaulich M, Lee YJ, Lo¨nn P, Springer AD, Meade BR, Dowdy SF. 2015. Efficient CRISPR-rAAV engineering ofendogenous genes to study protein function by allele-specific RNAi. Nucleic Acids Research 43:e45.DOI: https://doi.org/10.1093/nar/gku1403,PMID: 25586224Kim YJ, Wilson DM. 2012. Overview of base excision repair biochemistry. Current Molecular Pharmacology 5:3–13. DOI: https://doi.org/10.2174/1874467211205010003,PMID: 22122461Koike-Yusa H, Li Y, Tan EP, Velasco-Herrera MC, Yusa K. 2014. Genome-wide recessive genetic screening inmammalian cells with a lentiviral CRISPR-guide RNA library. Nature Biotechnology 32:267–273. DOI: https://doi.org/10.1038/nbt.2800,PMID: 24535568Kong H, Kucera RB, Jack WE. 1993. Characterization of a DNA polymerase from the hyperthermophile archaeathermococcus litoralis. vent DNA polymerase, steady state kinetics, thermal stability, Processivity, stranddisplacement, and exonuclease activities. The Journal of Biological Chemistry 268:1965–1975. PMID: 8420970Korkmaz G, Lopes R, Ugalde AP, Nevedomskaya E, Han R, Myacheva K, Zwart W, Elkon R, Agami R. 2016.Functional genetic screens for enhancer elements in the human genome using CRISPR-Cas9. NatureBiotechnology 34:192–198. DOI: https://doi.org/10.1038/nbt.3450,PMID: 26751173Kunkel TA. 1985. Rapid and efficient site-specific mutagenesis without phenotypic selection. PNAS 82:488–492.DOI: https://doi.org/10.1073/pnas.82.2.488,PMID: 3881765Kunkel TA. 2001. Oligonucleotide-directed mutagenesis without phenotypic selection. Current Protocols inNeuroscience Chapter 4:Unit 4.10. DOI: https://doi.org/10.1002/0471142301.ns0410s03,PMID: 18428474Li M, Brooks CL, Kon N, Gu W. 2004. A dynamic role of HAUSP in the p53-Mdm2 pathway. Molecular Cell 13:879–886. DOI: https://doi.org/10.1016/S1097-2765(04)00157-1,PMID: 15053880Li W, Xu H, Xiao T, Cong L, Love MI, Zhang F, Irizarry RA, Liu JS, Brown M, Liu XS. 2014. MAGeCK enablesrobust identification of essential genes from genome-scale CRISPR/Cas9 knockout screens. Genome Biology15:554. DOI: https://doi.org/10.1186/s13059-014-0554-4,PMID: 25476604Liberzon A, Subramanian A, Pinchback R, Thorvaldsdo´ ttir H, Tamayo P, Mesirov JP. 2011. Molecular signaturesdatabase (MSigDB) 3.0. Bioinformatics 27:1739–1740. DOI: https://doi.org/10.1093/bioinformatics/btr260,PMID: 21546393Makowski L, Soares A. 2003. Estimating the diversity of peptide populations from limited sequence data.Bioinformatics 19:483–489. DOI: https://doi.org/10.1093/bioinformatics/btg013,PMID: 12611803Martin M. 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal17:10. DOI: https://doi.org/10.14806/ej.17.1.200McIntyre GJ, Fanning GC. 2006. Design and cloning strategies for constructing shRNA expression vectors. BMCBiotechnology 6:1. DOI: https://doi.org/10.1186/1472-6750-6-1,PMID: 16396676Meitinger F, Anzola JV, Kaulich M, Richardson A, Stender JD, Benner C, Glass CK, Dowdy SF, Desai A, Shiau AK,Oegema K. 2016. 53bp1 and USP28 mediate p53 activation and G1 arrest after centrosome loss or extendedmitotic duration. The Journal of Cell Biology 214:155–166. DOI: https://doi.org/10.1083/jcb.201604081,PMID: 27432897Morgens DW, Deans RM, Li A, Bassik MC. 2016. Systematic comparison of CRISPR/Cas9 and RNAi screens foressential genes. Nature biotechnology 34:634–636. DOI: https://doi.org/10.1038/nbt.3567,PMID: 27159373Oliphant TE. 2010. Guide to NumPy. MIT press.Ong SH, Li Y, Koike-Yusa H, Yusa K. 2017. Optimised metrics for CRISPR-KO screens with second-generationgRNA libraries. Scientific Reports 7:7384. DOI: https://doi.org/10.1038/s41598-017-07827-z,PMID: 28785007Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 30 of 31Tools and resources Cell Biology Genetics and Genomics Pattanayak V, Lin S, Guilinger JP, Ma E, Doudna JA, Liu DR. 2013. High-throughput profiling of off-target DNAcleavage reveals RNA-programmed Cas9 nuclease specificity. Nature Biotechnology 31:839–843. DOI: https://doi.org/10.1038/nbt.2673,PMID: 23934178Potting C, Crochemore C, Moretti F, Nigsch F, Schmidt I, Manneville C, Carbone W, Knehr J, DeJesus R,Lindeman A, Maher R, Russ C, McAllister G, Reece-Hoyes JS, Hoffman GR, Roma G, Mu¨ller M, Sailer AW,Helliwell SB. 2018. Genome-wide CRISPR screen for PARKIN regulators reveals transcriptional repression as adeterminant of mitophagy. PNAS 115:E180–E189. DOI: https://doi.org/10.1073/pnas.1711023115Sanjana NE, Shalem O, Zhang F. 2014. Improved vectors and genome-wide libraries for CRISPR screening.Nature Methods 11:783–784. DOI: https://doi.org/10.1038/nmeth.3047,PMID: 25075903Sanjana NE, Wright J, Zheng K, Shalem O, Fontanillas P, Joung J, Cheng C, Regev A, Zhang F. 2016. High-resolution interrogation of functional elements in the noncoding genome. Science 353:1545–1549.DOI: https://doi.org/10.1126/science.aaf7613,PMID: 27708104Sanson KR, Hanna RE, Hegde M, Donovan KF, Strand C, Sullender ME, Vaimberg EW, Goodale A, Root DE,Piccioni F, Doench JG. 2018. Up, down, and out: next generation libraries for genome-wide CRISPRa, CRISPRi,and CRISPR-Cas9 knockout genetic screens. bioRxiv.DOI: https://doi.org/10.1101/356626Schmidt T, Schmid-Burgk JL, Hornung V. 2015. Synthesis of an arrayed sgRNA library targeting the humangenome. Scientific Reports 5:14987. DOI: https://doi.org/10.1038/srep14987,PMID: 26446710Shalem O, Sanjana NE, Hartenian E, Shi X, Scott DA, Mikkelson T, Heckl D, Ebert BL, Root DE, Doench JG,Zhang F. 2014. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science 343:84–87.DOI: https://doi.org/10.1126/science.1247005,PMID: 24336571Shapiro SS, Wilk MB. 1965. An analysis of variance test for normality (complete samples). Biometrika 52:591–611.DOI: https://doi.org/10.1093/biomet/52.3-4.591Sidhu SS. 2001. Engineering M13 for phage display. Biomolecular Engineering 18:57–63. DOI: https://doi.org/10.1016/S1389-0344(01)00087-9,PMID: 11535417Smith GP, Scott JK. 1993. Libraries of peptides and proteins displayed on filamentous phage. Methods inEnzymology 217:228–257. PMID: 7682645Spahn PN, Bath T, Weiss RJ, Kim J, Esko JD, Lewis NE, Harismendy O. 2017. PinAPL-Py: a comprehensive web-application for the analysis of CRISPR/Cas9 screens. Scientific Reports 7:15854. DOI: https://doi.org/10.1038/s41598-017-16193-9,PMID: 29158538Spencer DM, Bilardi RA, Koch TH, Post GC, Nafie JW, Kimura K, Cutts SM, Phillips DR. 2008. DNA repair inresponse to anthracycline-DNA adducts: a role for both homologous recombination and nucleotide excisionrepair. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis 638:110–121. DOI: https://doi.org/10.1016/j.mrfmmm.2007.09.005,PMID: 17961607Stewart SA, Dykxhoorn DM, Palliser D, Mizuno H, Yu EY, An DS, Sabatini DM, Chen IS, Hahn WC, Sharp PA,Weinberg RA, Novina CD. 2003. Lentivirus-delivered stable gene silencing by RNAi in primary cells. RNA 9:493–501. DOI: https://doi.org/10.1261/rna.2192803,PMID: 12649500Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR,Lander ES, Mesirov JP. 2005. Gene set enrichment analysis: a knowledge-based approach for interpretinggenome-wide expression profiles. Proceedings of the National Academy of Sciences 102:15545–15550.DOI: https://doi.org/10.1073/pnas.0506580102,PMID: 16199517Szklarczyk D, Morris JH, Cook H, Kuhn M, Wyder S, Simonovic M, Santos A, Doncheva NT, Roth A, Bork P,Jensen LJ, von Mering C. 2017. The STRING database in 2017: quality-controlled protein-protein associationnetworks, made broadly accessible. Nucleic Acids Research 45:D362–D368. DOI: https://doi.org/10.1093/nar/gkw937,PMID: 27924014van Brabant AJ, Stan R, Ellis NA. 2000. DNA helicases, genomic instability, and human genetic disease. AnnualReview of Genomics and Human Genetics 1:409–459. DOI: https://doi.org/10.1146/annurev.genom.1.1.409,PMID: 11701636Vidigal JA, Ventura A. 2015. Rapid and efficient one-step generation of paired gRNA CRISPR-Cas9 libraries.Nature Communications 6:8083. DOI: https://doi.org/10.1038/ncomms9083,PMID: 26278926Wang T, Wei JJ, Sabatini DM, Lander ES. 2014. Genetic screens in human cells using the CRISPR-Cas9 system.Science 343:80–84. DOI: https://doi.org/10.1126/science.1246981,PMID: 24336569Waskom M , Botvinnik O, Hobson P, Cole JB, Halchenko Y, Hoyer S, Mile A, Augspurger T, Yarkoni T, Megies T,Coelho LP, Wehner D, Ziegler E, Zaytsev YV, Hoppe T, Seabold S, Cloud P, Koskinen M, Meyer K, Qalieh A,et al. 2014. seaborn. Zenodo. v0.9.0.Weinhold N, Jacobsen A, Schultz N, Sander C, Lee W. 2014. Genome-wide analysis of noncoding regulatorymutations in cancer. Nature Genetics 46:1160–1165. DOI: https://doi.org/10.1038/ng.3101,PMID: 25261935Wyvekens N, Topkar VV, Khayter C, Joung JK, Tsai SQ. 2015. Dimeric CRISPR RNA-Guided FokI-dCas9nucleases directed by truncated gRNAs for highly specific genome editing. Human Gene Therapy 26:425–431.DOI: https://doi.org/10.1089/hum.2015.084,PMID: 26068112Zerbino DR, Achuthan P, Akanni W, Amode MR, Barrell D, Bhai J, Billis K, Cummins C, Gall A, Giro´ n CG, Gil L,Gordon L, Haggerty L, Haskell E, Hourlier T, Izuogu OG, Janacek SH, Juettemann T, To JK, Laird MR, et al.2018. Ensembl 2018. Nucleic Acids Research 46:D754–D761. DOI: https://doi.org/10.1093/nar/gkx1098,PMID: 29155950Wegner et al. eLife 2019;8:e42549. DOI: https://doi.org/10.7554/eLife.42549 31 of 31Tools and resources Cell Biology Genetics and GenomicsCitations (14)References (87)... The protocol for 3Cs multiplex-DNA synthesis was adapted from Wegner et al. (2019) and optimized for reactions on the 3Cs multiplex template plasmid with two specific annealing sites (29,30). The oligonucleotides that were used for 3Cs reactions and the suppliers are listed separately (Supplementary Table S1). ...... The protocol for 3Cs multiplex-DNA synthesis was adapted from Wegner et al. (2019) and optimized for reactions on the 3Cs multiplex template plasmid with two specific annealing sites (29,30). The oligonucleotides that were used for 3Cs reactions and the suppliers are listed separately (Supplementary Table S1). ...... Cell culture work was performed as described previously (29). In brief, HEK293T cells (ATCC, CRL-3216) were maintained in Dulbecco s modified Eagle s medium (DMEM, Thermo Fisher Scientific) and puromycinsensitive human telomerase-immortalized retinal pigmented epithelial (RPE1) cells (provided by Andrew Holland) in DMEM: Nutrient Mixture F-12 (DMEM/F12, Thermo Fisher Scientific), each supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific) and 1% penicillin-streptomycin (Sigma-Aldrich) at 37 • C with 5% CO 2 . ...Minimized combinatorial CRISPR screens identify genetic interactions in autophagyArticleFull-text availableApr 2021Nucleic Acids Res Valentina Diehl Martin Wegner Paolo Grumati Manuel KaulichCombinatorial CRISPR-Cas screens have advanced the mapping of genetic interactions, but their experimental scale limits the number of targetable gene combinations. Here, we describe 3Cs mul-tiplexing, a rapid and scalable method to generate highly diverse and uniformly distributed com-binatorial CRISPR libraries. We demonstrate that the library distribution skew is the critical determinant of its required screening coverage. By circumventing iterative cloning of PCR-amplified oligonu-cleotides, 3Cs multiplexing facilitates the generation of combinatorial CRISPR libraries with low distribution skews. We show that combinatorial 3Cs libraries can be screened with minimal coverages, reducing associated efforts and costs at least 10-fold. We apply a 3Cs multiplexing library targeting 12,736 autophagy gene combinations with 247,032 paired gRNAs in viability and reporter-based enrichment screens. In the viability screen, we identify, among others, the synthetic lethal WDR45B-PIK3R4 and the proliferation-enhancing ATG7-KEAP1 genetic interactions. In the reporter-based screen, we identify over 1,570 essential genetic interactions for autophagy flux, including interactions among par-alogous genes, namely ATG2A-ATG2B, GABARAP-MAP1LC3B and GABARAP-GABARAPL2. However, we only observe few genetic interactions within par-alogous gene families of more than two members, indicating functional compensation between them. This work establishes 3Cs multiplexing as a platform for genetic interaction screens at scale.ViewShow abstract... The protocol for 3Cs multiplex-DNA synthesis was adapted from Wegner et al. (2019) and optimized for reactions on the 3Cs multiplex template plasmid with two specific annealing sites (29,30). The oligonucleotides that were used for 3Cs reactions and the suppliers are listed separately (Supplementary Table S1). ...... The protocol for 3Cs multiplex-DNA synthesis was adapted from Wegner et al. (2019) and optimized for reactions on the 3Cs multiplex template plasmid with two specific annealing sites (29,30). The oligonucleotides that were used for 3Cs reactions and the suppliers are listed separately (Supplementary Table S1). ...... Cell culture work was performed as described previously (29). In brief, HEK293T cells (ATCC, CRL-3216) were maintained in Dulbecco s modified Eagle s medium (DMEM, Thermo Fisher Scientific) and puromycinsensitive human telomerase-immortalized retinal pigmented epithelial (RPE1) cells (provided by Andrew Holland) in DMEM: Nutrient Mixture F-12 (DMEM/F12, Thermo Fisher Scientific), each supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific) and 1% penicillin-streptomycin (Sigma-Aldrich) at 37 • C with 5% CO 2 . ...Minimized combinatorial CRISPR screens identify genetic interactions in autophagyArticleFull-text availableMay 2021Nucleic Acids ResIbrahim Halil PolatCombinatorial CRISPR-Cas screens have advanced the mapping of genetic interactions, but their experimental scale limits the number of targetable gene combinations. Here, we describe 3Cs multiplexing, a rapid and scalable method to generate highly diverse and uniformly distributed combinatorial CRISPR libraries. We demonstrate that the library distribution skew is the critical determinant of its required screening coverage. By circumventing iterative cloning of PCR-amplified oligonucleotides, 3Cs multiplexing facilitates the generation of combinatorial CRISPR libraries with low distribution skews. We show that combinatorial 3Cs libraries can be screened with minimal coverages, reducing associated efforts and costs at least 10-fold. We apply a 3Cs multiplexing library targeting 12,736 autophagy gene combinations with 247,032 paired gRNAs in viability and reporter-based enrichment screens. In the viability screen, we identify, among others, the synthetic lethal WDR45B-PIK3R4 and the proliferation-enhancing ATG7-KEAP1 genetic interactions. In the reporter-based screen, we identify over 1,570 essential genetic interactions for autophagy flux, including interactions among paralogous genes, namely ATG2A-ATG2B, GABARAP-MAP1LC3B and GABARAP-GABARAPL2. However, we only observe few genetic interactions within paralogous gene families of more than two members, indicating functional compensation between them. This work establishes 3Cs multiplexing as a platform for genetic interaction screens at scale.ViewShow abstract... Since approaches to perform highly efficient KI have been reviewed before, we refer the interested reader to these references [39,40]. The ease of directing the CRISPR/Cas system to almost any region in a given genome makes it a well-suited system for high-throughput functional genomic studies for the unbiased correlation of genotypes and phenotypes [41]. ...... However, at least one significantly enriched site was more than 250kb away from the CDKN1A TSS, suggesting a role of non-protein-coding sequence elements in doxorubicin resistance. In line with this, Wegner et al. generated the largest to-date CRISPR library targeting coding and non-coding regions based on Cas9′s gRNA-sequence preferences and applied the reagent to doxorubicin-exposed hTERT-immortalized RPE1 cells [41]. The Leukotriene C4 G-protein-coupled eicosanoid receptor CysLTR2, that was previously reported to induce doxorubicin resistance by preventing the accumulation of reactive oxygen species [136], was the strongest protein-coding hit gene. ...... However, 50.3% of retrieved gRNAs accounted for non-protein-coding regions with eight gRNAs targeting positions of the a-kinase anchor protein AKAP6 and the aspartoacylase ASPA2 genes, of which AKAP6 was previously linked to doxorubicin sensitivity [137]. Moreover, Wegner et al. identified predicted promoter and enhancer sequences, lincRNAs, pseudogenes, as well as CTCF binding sites, further supporting a hitherto underappreciated role of noncoding sequences in doxorubicin resistance [41]. ...Functional Genomics Approaches to Elucidate Vulnerabilities of Intrinsic and Acquired Chemotherapy ResistanceArticleFull-text availableJan 2021 Ronay Çetin Eva Quandt Manuel KaulichDrug resistance is a commonly unavoidable consequence of cancer treatment that results in therapy failure and disease relapse. Intrinsic (pre-existing) or acquired resistance mechanisms can be drug-specific or be applicable to multiple drugs, resulting in multidrug resistance. The presence of drug resistance is, however, tightly coupled to changes in cellular homeostasis, which can lead to resistance-coupled vulnerabilities. Unbiased gene perturbations through RNAi and CRISPR technologies are invaluable tools to establish genotype-to-phenotype relationships at the genome scale. Moreover, their application to cancer cell lines can uncover new vulnerabilities that are associated with resistance mechanisms. Here, we discuss targeted and unbiased RNAi and CRISPR efforts in the discovery of drug resistance mechanisms by focusing on first-in-line chemotherapy and their enforced vulnerabilities, and we present a view forward on which measures should be taken to accelerate their clinical translation.ViewShow abstract... This has vastly contributed to the increase in screening reproducibility [16,67,108,153]. The majority of reported CRISPR screens were designed to identify novel gene to phenotype relationships, while only few studies investigated functional aspects of non-coding sequences [23,134,160]. Strategies to efficiently interrogate and functionally annotate non-coding regions apply either DNA tiling or excision approaches. For both approaches, genomic regions are targeted by all possible gRNAs within a given region or the DNA connecting two juxtaposed gRNA-target sites is excised and lost, respectively [23,76,126]. ...DGK and DZHK position paper on genome editing: basic science applications and future perspectiveArticleFull-text availableDec 2021Basic Res Cardiol Ralf P BrandesAnne Dueck Stefan EngelhardtWolfgang WurstFor a long time, gene editing had been a scientific concept, which was limited to a few applications. With recent developments, following the discovery of TALEN zinc-finger endonucleases and in particular the CRISPR/Cas system, gene editing has become a technique applicable in most laboratories. The current gain- and loss-of function models in basic science are revolutionary as they allow unbiased screens of unprecedented depth and complexity and rapid development of transgenic animals. Modifications of CRISPR/Cas have been developed to precisely interrogate epigenetic regulation or to visualize DNA complexes. Moreover, gene editing as a clinical treatment option is rapidly developing with first trials on the way. This article reviews the most recent progress in the field, covering expert opinions gathered during joint conferences on genome editing of the German Cardiac Society (DGK) and the German Center for Cardiovascular Research (DZHK). Particularly focusing on the translational aspect and the combination of cellular and animal applications, the authors aim to provide direction for the development of the field and the most frequent applications with their problems.ViewShow abstract... Here, we use a genome-wide CRISPR perturbation library consisting of partially randomized degenerated oligonucleotides (5 -NNDNNNNNHNNNNHDHNVVR-3 ) with flanking 3Cs homology regions which was created using ssDNA of template-plasmids and site-specific mutagenesis targeting coding and non-coding regions of the human genome in hTERT-RPE1 cells from ATCC (CRL-4000) [3]. Genomic coordiantes of the gRNAs have been obtained from this work (see Supp. ...Computational prediction of CRISPR-impaired non-coding regulatory regionsPreprintFull-text availableDec 2020Nina Baumgarten Florian Schmidt Martin Wegner Marcel SchulzGenome-wide CRISPR screens are becoming more widespread and allow the simultaneous interrogation of thousands of genomic regions. Although recent progress has been made in the analysis of CRISPR screens, it is still an open problem how to interpret CRISPR mutations in non-coding regions of the genome. Most of the tools concentrate on the interpretation of mutations introduced in gene coding regions. We introduce a computational pipeline that uses epigenomic information about regulatory elements for the interpretation of CRISPR mutations in non-coding regions. We illustrate our approach on the analysis of a genome-wide CRISPR screen in hTERT-RPE-1 cells and reveal novel regulatory elements that mediate chemoresistance against doxorubicin in these cells. We infer links to established and to novel chemoresistance genes. Our approach is general and can be applied on any cell type and with different CRISPR enzymes.ViewShow abstract... A further example for a recently emerged CRISPR/Cas9 application is parallel targeting of many loci with gRNA libraries, required for instance when targeting transcription factor binding sites (TFBSs) or their neighborhoods (Shariati et al, 2019). Other screening-oriented CRISPR/Cas9-based applications include genomewide visualization or complex gRNA libraries to investigate cell fitness (Wegner et al, 2019). For such applications, the total number of gRNAs, or library complexity, directly correlates with effort and costs. ...multicrispr: gRNA design for prime editing and parallel targeting of thousands of targetsArticleFull-text availableNov 2020 Aditya Bhagwat Johannes GraumannRené Wiegandt Mario LoosoTargeting the coding genome to introduce nucleotide deletions/insertions via the CRISPR/Cas9 technology has become a standard procedure. It has quickly spawned a multitude of methods such as prime editing, APEX proximity labeling, or homology directed repair, for which supporting bioinformatics tools are, however, lagging behind. New CRISPR/Cas9 applications often require specific gRNA design functionality, and a generic tool is critically missing. Here, we introduce multicrispr, an R/bioconductor tool, intended to design individual gRNAs and complex gRNA libraries. The package is easy to use; detects, scores, and filters gRNAs on both efficiency and specificity; visualizes and aggregates results per target or CRISPR/Cas9 sequence; and finally returns both genomic ranges and sequences of gRNAs. To be generic, multicrispr defines and implements a genomic arithmetic framework as a basis for facile adaptation to techniques recently introduced such as prime editing or yet to arise. Its performance and design concepts such as target set–specific filtering render multicrispr a tool of choice when dealing with screening-like approaches.ViewShow abstract... . Other screening oriented Crispr/Cas9-based applications include genome wide visualization , or complex gRNA libraries to investigate cell fitness (Wegner et al. 2019). For such applications, the total number of gRNAs, or library complexity, directly correlates with effort and costs. ...multicrispr: fast gRNA designer enables prime editing and parallel targeting of thousands of targetsPreprintFull-text availableApr 2020 Aditya Bhagwat Johannes GraumannRené Wiegandt Mario LoosoTargeting the coding genome to introduce single nucleotide deletions/insertions via Crispr/Cas9 technology has become a standard procedure in recent years. Due to the whirlwind pace of evolution of Crispr/Cas9 based methods for which Prime editing, Crispr/Cas9 assisted APEX proximity labeling of proteins, or homology directed repair (HDR) are just innovative recent examples, supporting bioinformatic tools are, however, lagging behind. New methods often require specific guide-RNA (gRNA) design functionality, and a generic gRNA design tool is critically missing. Here we review gRNA designer software and introduce multicrispr, an R based tool intended to design individual gRNAs as well as gRNA libraries targeting a multitude of genomic loci in parallel. The package is easy to use, it detects, scores and filters gRNAs on both efficiency and specificity, it visualizes and aggregates results per target or Crispr/Cas9 sequence, and finally returns both genomic ranges as well as sequences of preferred, off target-free gRNAs. In order to be generic, multicrispr defines and implements a genomic arithmetics framework as a basis for facile adaptation to techniques yet to arise. Its performance and new gRNA design concepts such as target set specific filtering for gRNA libraries render multicrispr the tool of choice when dealing with screening-like approaches.ViewShow abstract... We hypothesize that the broadening of library distribution is due to sequence-specific differences in synthesis or amplification efficiency. A recently published approach to synthesize covalently-closed-circular-synthesized (3Cs) gRNA libraries may thus be a promising technology for substantial reduction of library width and experiment size [52]. ...Gscreend: Modelling asymmetric count ratios in CRISPR screens to decrease experiment size and improve phenotype detectionArticleFull-text availableMar 2020GENOME BIOL Katharina ImkellerGiulia AmbrosiMichael BoutrosWolfgang HuberPooled CRISPR screens are a powerful tool to probe genotype-phenotype relationships at genome-wide scale. However, criteria for optimal design are missing, and it remains unclear how experimental parameters affect results. Here, we report that random decreases in gRNA abundance are more likely than increases due to bottle-neck effects during the cell proliferation phase. Failure to consider this asymmetry leads to loss of detection power. We provide a new statistical test that addresses this problem and improves hit detection at reduced experiment size. The method is implemented in the R package gscreend, which is available at http://bioconductor.org/packages/gscreend.ViewShow abstractComputational prediction of CRISPR-impaired non-coding regulatory regionsArticleMar 2021Biol ChemNina Baumgarten Florian Schmidt Martin Wegner Marcel SchulzGenome-wide CRISPR screens are becoming more widespread and allow the simultaneous interrogation of thousands of genomic regions. Although recent progress has been made in the analysis of CRISPR screens, it is still an open problem how to interpret CRISPR mutations in non-coding regions of the genome. Most of the tools concentrate on the interpretation of mutations introduced in gene coding regions. We introduce a computational pipeline that uses epigenomic information about regulatory elements for the interpretation of CRISPR mutations in non-coding regions. We illustrate our analysis protocol on the analysis of a genome-wide CRISPR screen in hTERT-RPE1 cells and reveal novel regulatory elements that mediate chemoresistance against doxorubicin in these cells. We infer links to established and to novel chemoresistance genes. Our analysis protocol is general and can be applied on any cell type and with different CRISPR enzymes.ViewShow abstractCombinatorial CRISPR screening reveals functional buffering in autophagyPreprintFull-text availableJul 2020 Valentina Diehl Martin Wegner Paolo Grumati Manuel KaulichFunctional genomics studies in model organisms and human cell lines provided important insights into gene functions and their context-dependent role in genetic circuits. However, our functional understanding of many of these genes and how they combinatorically regulate key biological processes, remains limited. To enable the SpCas9-dependent mapping of gene-gene interactions in human cells, we established 3Cs multiplexing for the generation of combinatorial gRNA libraries in a distribution-unbiased manner and demonstrate its robust performance. The optimal number for combinatorial hit calling was 16 gRNA pairs and the skew of a library′s distribution was identified as a critical parameter dictating experimental scale and data quality. Our approach enabled us to investigate 247,032 gRNA-pairs targeting 12,736 gene-interactions in human autophagy. We identified novel genes essential for autophagy and provide experimental evidence that gene-associated categories of phenotypic strengths exist in autophagy. Furthermore, circuits of autophagy gene interactions reveal redundant nodes driven by paralog genes. Our combinatorial 3Cs approach is broadly suitable to investigate unexpected gene-interaction phenotypes in unperturbed and diseased cell contexts.ViewShow abstractShow moreUp, down, and out: next generation libraries for genome-wide CRISPRa, CRISPRi, and CRISPR-Cas9 knockout genetic screensPreprintFull-text availableJun 2018Kendall R. SansonRuth E. Hanna Mudra HegdeJohn G. DoenchAdvances in CRISPR-Cas9 technology have enabled the flexible modulation of gene expression at large scale. In particular, the creation of genome-wide libraries for CRISPR knockout (CRISPRko), CRISPR interference (CRISPRi), and CRISPR activation (CRISPRa) has allowed gene function to be systematically interrogated. Here, we evaluate numerous CRISPRko libraries and show that our recently-described CRISPRko library (Brunello) is more effective than previously published libraries at distinguishing essential and non-essential genes, providing approximately the same perturbation-level performance improvement over GeCKO libraries as GeCKO provided over RNAi. Additionally, we developed genome-wide libraries for CRISPRi (Dolcetto) and CRISPRa (Calabrese). Negative selection screens showed that Dolcetto substantially outperforms existing CRISPRi libraries with fewer sgRNAs per gene and achieves comparable performance to CRISPRko in the detection of gold-standard essential genes. We also conducted positive selection CRISPRa screens and show that Calabrese outperforms the SAM library approach at detecting vemurafenib resistance genes. We further compare CRISPRa to genome-scale libraries of open reading frames (ORFs). Together, these libraries represent a suite of genome-wide tools to efficiently interrogate gene function with multiple modalities.ViewShow abstractCRISPR–Cas9 genome editing induces a p53-mediated DNA damage responseArticleFull-text availableJul 2018Nat Med Emma HaapaniemiSandeep Botla Jenna PerssonJussi TaipaleHere, we report that genome editing by CRISPR-Cas9 induces a p53-mediated DNA damage response and cell cycle arrest in immortalized human retinal pigment epithelial cells, leading to a selection against cells with a functional p53 pathway. Inhibition of p53 prevents the damage response and increases the rate of homologous recombination from a donor template. These results suggest that p53 inhibition may improve the efficiency of genome editing of untransformed cells and that p53 function should be monitored when developing cell-based therapies utilizing CRISPR-Cas9.ViewShow abstractGenome-wide CRISPR screen for PARKIN regulators reveals transcriptional repression as a determinant of mitophagyArticleFull-text availableDec 2017P NATL ACAD SCI USAChristoph PottingChristophe Crochemore Francesca Moretti Stephen B HelliwellPARKIN, an E3 ligase mutated in familial Parkinson s disease, promotes mitophagy by ubiquitinating mitochondrial proteins for efficient engagement of the autophagy machinery. Specifically, PARKIN-synthesized ubiquitin chains represent targets for the PINK1 kinase generating phosphoS65-ubiquitin (pUb), which constitutes the mitophagy signal. Physiological regulation of PARKIN abundance, however, and the impact on pUb accumulation are poorly understood. Using cells designed to discover physiological regulators of PARKIN abundance, we performed a pooled genome-wide CRISPR/Cas9 knockout screen. Testing identified genes individually resulted in a list of 53 positive and negative regulators. A transcriptional repressor network including THAP11 was identified and negatively regulates endogenous PARKIN abundance. RNAseq analysis revealed the PARKIN-encoding locus as a prime THAP11 target, and THAP11 CRISPR knockout in multiple cell types enhanced pUb accumulation. Thus, our work demonstrates the critical role of PARKIN abundance, identifies regulating genes, and reveals a link between transcriptional repression and mitophagy, which is also apparent in human induced pluripotent stem cell-derived neurons, a disease-relevant cell type.ViewShow abstractEnsembl 2018ArticleFull-text availableNov 2017Nucleic Acids ResDaniel Zerbino Premanand Achuthan Wasiu A. AkanniPaul FlicekThe Ensembl project has been aggregating, processing, integrating and redistributing genomic datasets since the initial releases of the draft human genome, with the aim of accelerating genomics research through rapid open distribution of public data. Large amounts of raw data are thus transformed into knowledge, which is made available via a multitude of channels, in particular our browser (http://www.ensembl.org). Over time, we have expanded in multiple directions. First, our resources describe multiple fields of genomics, in particular gene annotation, comparative genomics, genetics and epigenomics. Second, we cover a growing number of genome assemblies; Ensembl Release 90 contains exactly 100. Third, our databases feed simultaneously into an array of services designed around different use cases, ranging from quick browsing to genome-wide bioinformatic analysis. We present here the latest developments of the Ensembl project, with a focus on managing an increasing number of assemblies, supporting efforts in genome interpretation and improving our browser.ViewShow abstractPinAPL-Py: A comprehensive web-application for the analysis of CRISPR/Cas9 screensArticleFull-text availableNov 2017 Philipp SpahnTyler Bath Ryan Joseph WeissOlivier HarismendyLarge-scale genetic screens using CRISPR/Cas9 technology have emerged as a major tool for functional genomics. With its increased popularity, experimental biologists frequently acquire large sequencing datasets for which they often do not have an easy analysis option. While a few bioinformatic tools have been developed for this purpose, their utility is still hindered either due to limited functionality or the requirement of bioinformatic expertise. To make sequencing data analysis of CRISPR/Cas9 screens more accessible to a wide range of scientists, we developed a Platform-independent Analysis of Pooled Screens using Python (PinAPL-Py), which is operated as an intuitive web-service. PinAPL-Py implements state-of-the-art tools and statistical models, assembled in a comprehensive workflow covering sequence quality control, automated sgRNA sequence extraction, alignment, sgRNA enrichment/depletion analysis and gene ranking. The workflow is set up to use a variety of popular sgRNA libraries as well as custom libraries that can be easily uploaded. Various analysis options are offered, suitable to analyze a large variety of CRISPR/Cas9 screening experiments. Analysis output includes ranked lists of sgRNAs and genes, and publication-ready plots. PinAPL-Py helps to advance genome-wide screening efforts by combining comprehensive functionality with user-friendly implementation. PinAPL-Py is freely accessible at http://pinapl-py.ucsd.edu with instructions and test datasets.ViewShow abstractOptimised metrics for CRISPR-KO screens with second-generation gRNA librariesArticleFull-text availableDec 2017Swee Hoe Ong Yilong LiHiroko Koike-YusaKosuke YusaGenome-wide CRISPR-based knockout (CRISPR-KO) screening is an emerging technique which enables systematic genetic analysis of a cellular or molecular phenotype in question. Continuous improvements, such as modifications to the guide RNA (gRNA) scaffold and the development of gRNA on-target prediction algorithms, have since been made to increase their screening performance. We compared the performance of three available second-generation human genome-wide CRISPR-KO libraries that included at least one of the improvements, and examined the effect of gRNA scaffold, number of gRNAs per gene and number of replicates on screen performance. We identified duplicated screens using a library with 6 gRNAs per gene as providing the best trade-off. Despite the improvements, we found that each improved library still has library-specific false negatives and, for the first time, estimated the false negative rates of CRISPR-KO screens, which are between 10% and 20%. Our newly-defined optimal screening parameters would be helpful in designing screens and constructing bespoke gRNA libraries.ViewShow abstractThe STRING database in 2017: quality-controlled protein-protein association networks, made broadly accessibleArticleFull-text availableOct 2016Nucleic Acids Res Damian Szklarczyk John H Morris Helen Cook Christian von MeringA system-wide understanding of cellular function requires knowledge of all functional interactions between the expressed proteins. The STRING database aims to collect and integrate this information, by consolidating known and predicted protein-protein association data for a large number of organisms. The associations in STRING include direct (physical) interactions, as well as indirect (functional) interactions, as long as both are specific and biologically meaningful. Apart from collecting and reassessing available experimental data on protein-protein interactions, and importing known pathways and protein complexes from curated databases, interaction predictions are derived from the following sources: (i) systematic co-expression analysis, (ii) detection of shared selective signals across genomes, (iii) automated text-mining of the scientific literature and (iv) computational transfer of interaction knowledge between organisms based on gene orthology. In the latest version 10.5 of STRING, the biggest changes are concerned with data dissemination: the web frontend has been completely redesigned to reduce dependency on outdated browser technologies, and the database can now also be queried from inside the popular Cytoscape software framework. Further improvements include automated background analysis of user inputs for functional enrichments, and streamlined download options. The STRING resource is available online, at http://string-db.org/.ViewShow abstractGenome-scale CRISPR-Cas9 knockout and transcriptional activation screeningArticleMar 2017NAT PROTOC Julia Joung Silvana Konermann Jonathan GootenbergFeng ZhangForward genetic screens are powerful tools for the unbiased discovery and functional characterization of specific genetic elements associated with a phenotype of interest. Recently, the RNA-guided endonuclease Cas9 from the microbial CRISPR (clustered regularly interspaced short palindromic repeats) immune system has been adapted for genome-scale screening by combining Cas9 with pooled guide RNA libraries. Here we describe a protocol for genome-scale knockout and transcriptional activation screening using the CRISPR-Cas9 system. Custom- or ready-made guide RNA libraries are constructed and packaged into lentiviral vectors for delivery into cells for screening. As each screen is unique, we provide guidelines for determining screening parameters and maintaining sufficient coverage. To validate candidate genes identified by the screen, we further describe strategies for confirming the screening phenotype, as well as genetic perturbation, through analysis of indel rate and transcriptional activation. Beginning with library design, a genome-scale screen can be completed in 9–15 weeks, followed by 4–5 weeks of validation.ViewShow abstractThe BLUEPRINT Data Analysis PortalArticleNov 2016 José María Fernández Victor de la TorreDavid Richardson Alfonso ValenciaThe impact of large and complex epigenomic datasets on biological insights or clinical applications is limited by the lack of accessibility by easy, intuitive, and fast tools. Here, we describe an epigenomics comparative cyber-infrastructure (EPICO), an open-access reference set of libraries to develop comparative epigenomic data portals. Using EPICO, large epigenome projects can make available their rich datasets to the community without requiring specific technical skills. As a first instance of EPICO, we implemented the BLUEPRINT Data Analysis Portal (BDAP). BDAP provides a desktop for the comparative analysis of epigenomes of hematopoietic cell types based on results, such as the position of epigenetic features, from basic analysis pipelines. The BDAP interface facilitates interactive exploration of genomic regions, genes, and pathways in the context of differentiation of hematopoietic lineages. This work represents initial steps toward broadly accessible integrative analysis of epigenomic data across international consortia. EPICO can be accessed at https://github.com/inab, and BDAP can be accessed at http://blueprint-data.bsc.es.ViewShow abstractHigh-resolution interrogation of functional elements in the noncoding genomeArticleSep 2016SCIENCENeville E. Sanjana Jason WrightKaijie ZhengFeng ZhangThe noncoding genome affects gene regulation and disease, yet we lack tools for rapid identification and manipulation of noncoding elements. We developed a CRISPR screen using ∼18,000 single guide RNAs targeting 700 kilobases surrounding the genes NF1, NF2, and CUL3, which are involved in BRAF inhibitor resistance in melanoma. We find that noncoding locations that modulate drug resistance also harbor predictive hallmarks of noncoding function. With a subset of regions at the CUL3 locus, we demonstrate that engineered mutations alter transcription factor occupancy and long-range and local epigenetic environments, implicating these sites in gene regulation and chemotherapeutic resistance. Through our expansion of the potential of pooled CRISPR screens, we provide tools for genomic discovery and for elucidating biologically relevant mechanisms of gene regulation. © 2016, American Association for the Advancement of Science. All rights reserved.ViewShow abstractShow moreAdvertisementRecommendationsDiscover moreProjectMapping the Landscape of Genetic Dependencies in Cancer Manuel KaulichKnowing the landscape of genetic vulnerabilities across cancer types is important for a deeper understanding of cancer biology, the discovery of new targets for drug development, overcoming treatme nt resistance and tailoring therapies to specific mutational contexts for precision medicine. The development of comprehensive and tumor-specific maps of genetic dependencies will require international efforts and coordination. ... [more]View projectPreprintFull-text availableCRISPR/Cas9-targeted removal of unwanted sequences from small-RNA sequencing librariesMay 2018 Andrew Hardigan Brian Scott RobertsDianna E. Moore[...] Richard M MyersIn small RNA (smRNAs) sequencing studies, highly abundant molecules such as adapter dimer products and tissue-specific microRNAs (miRNAs) inhibit accurate quantification of lowly expressed species. We previously developed a method to selectively deplete highly abundant miRNAs. However, this method does not deplete adapter dimer ligation products that, unless removed by gel-separation, comprise ... [Show full abstract] most of the library. Here, we have adapted and modified recently described methods for CRISPR/Cas9-based Depletion of Abundant Species by Hybridization (DASH) to smRNA-seq, which we have termed miRNA and Adapter Dimer - DASH (MAD-DASH). In MAD-DASH, Cas9 is complexed with sgRNAs targeting adapter dimer ligation products, alongside highly expressed tissue-specific smRNAs, for cleavage in vitro. This process dramatically reduces ( 90%) adapter dimer and targeted smRNA sequences, is multiplexable, shows minimal off-target effects, improves the quantification of lowly expressed miRNAs from human plasma and tissue derived RNA, and obviates the need for gel-separation, greatly increasing sample throughput. Additionally, the method is fully customizable to other smRNA-seq preparation methods. Like depletion of ribosomal RNA for mRNA-seq and mitochondrial DNA for ATAC-seq, our method allows for greater proportional read-depth of non-targeted sequences.View full-textArticleFull-text availableTCR/CD3 mediated stop-signal is decoupled in T-cells from Ctla4 deficient miceFebruary 2008 · Immunology LettersJos Downey Andrew M Smith Helga Schneider[...] Christopher RuddCTLA-4 is a co-receptor that plays a pivotal role in regulating the threshold for T-cell activation. We recently reported that CTLA-4 ligation can over-ride the stop-signal induced by anti-CD3 ligation [Schneider H, Downey J, Smith A, Zinselmeyer BH, Rush C, Brewer JM, et al. Reversal of the TCR stop-signal by CTLA-4. Science 2006;313:1972]. While these studies compared CTLA-4 positive and ... [Show full abstract] negative T-cells from normal mice, little is known regarding the behaviour of T-cells from diseased Ctla4 deficient mice with auto-proliferative disease. In this study, we show that while activated wild-type and Ctla-4-/- T-cells have similar rates of motility, Ctla-4-/- T-cells show a marked resistance to the induction of a stop-signal by anti-CD3 ligation. By contrast, T-cells from normal mice and CD28 deficient mice underwent a normal slowing of motility in response to anti-CD3 ligation. Our findings identify a fundamental difference between normal versus CTLA-4-/- T-cells from diseased mice in the regulation of motility by anti-CD3 ligation. This dysregulation of motility may contribute to the tissue infiltration and the autoimmune disorder observed in Ctla-4-/- mice.View full-textArticleQuantitative analysis of the T cell receptor signaling network in response to altered peptide ligand...November 2006Lucia WilleUnderstanding the adaptive immune system poses a great conceptual challenge. It has evolved to respond to foreign invaders with exquisite sensitivity and selectivity. In particular, the T cell branch of the immune system is trained to distinguish between self and non-self. This requires that a single receptor, the T cell receptor, bind to multiple ligands resulting in different cell fates, based ... [Show full abstract] in part on the avidity of the ligand. To address the question of ligand affinity discrimination in T cells, several T cell lines, both mouse and human, were screened for their ability to exhibit multiple cell fates in response to stimulation through the T cell receptor. A hybridoma system was identified that exhibits different levels of both apoptosis and cytokine production in response to three altered peptide ligands. We investigated how the consequent downstream signaling networks integrate to ultimately govern avidity-appropriate T cell responses in this hybridoma system. Here, we hypothesized that a quantitative combination of key downstream network signals can effectively represent the information processing generated by TCR ligation, providing a model capable of interpreting and predicting T cell functional responses.Read moreArticle[Examination of the hepatic artery in rats following its ligation and the effect of this ligation on...April 1974 · Zentralblatt für Veterinärmedizin. Reihe C: Anatomie, Histologie, Embryologie Helmut SinzingerH UnterbergerRead moreArticleMiniSAGE: Gene Expression Profiling Using Serial Analysis of Gene Expression from 1 μg Total RNAJanuary 2001 · Analytical Biochemistry Shui YeLi Q Zhang Fang Zheng[...] Peter O KwiterovichThe use of serial analysis of gene expression (SAGE) to determine gene expression profiles is increasing because the technique can provide absolute transcript numbers in a digital format and identify new genes. We developed a miniSAGE technique, which uses only 1 μg total RNA and reduces the amount of the starting material by 250- to 500-fold. Unlike the other modified SAGE methods, the miniSAGE ... [Show full abstract] technique does not require the additional PCR amplifications. The additional PCR amplifications potentially introduce bias and compromise the quantitative aspects of the SAGE method. Three key modifications in the miniSAGE technique are: (i) using the phase lock gel (PLG, Eppendorf) to increase the recovery and the purity of DNA material after each phenol extraction step; (ii) reducing the amount of linkers in the ligation, thereby minimizing their interference with SAGE ditag amplification and increasing the SAGE ditag yield; and (iii) employing the mRNA capture kit (Boehringer Mannheim) to allow the first five steps: mRNA isolation, cDNA synthesis, enzyme cleavage of cDNA, binding of the cleaved biotin–cDNA to the streptavidin-magnetic beads, ligating linkers to the bound cDNA, and the release of cDNA tags to occur within one tube to significantly reduce the loss of material between successive steps. Two fibroblast SAGE libraries have been successfully prepared. The preliminary analysis of 3838 tags from one library demonstrated a typical fibroblast gene expression pattern. This miniSAGE technique will permit a broader application of SAGE.Read moreLast Updated: 08 May 2021Discover the world s researchJoin ResearchGate to find the people and research you need to help your work.Join for free ResearchGate iOS AppGet it from the App Store now.InstallKeep up with your stats and moreAccess scientific knowledge from anywhere orDiscover by subject areaRecruit researchersJoin for freeLoginEmail Tip: Most researchers use their institutional email address as their ResearchGate loginPasswordForgot password? Keep me logged inLog inorContinue with GoogleWelcome back! Please log in.Email · HintTip: Most researchers use their institutional email address as their ResearchGate loginPasswordForgot password? Keep me logged inLog inorContinue with GoogleNo account? Sign upCompanyAbout usNewsCareersSupportHelp CenterBusiness solutionsAdvertisingRecruiting© 2008-2021 ResearchGate GmbH. All rights reserved.TermsPrivacyCopyrightImprint

>>> 更多资讯详情请访问蚂蚁淘商城