Summary
Perturbing expression is a powerful way to understand the role of individual genes, but can be challenging in important models. CRISPR-Cas screens in human induced pluripotent stem cells (iPSCs) are of limited efficiency due to DNA break-induced stress, while the less stressful silencing with an inactive Cas9 has been considered less effective so far. Here, we developed the dCas9-KRAB-MeCP2 fusion protein for screening in iPSCs from multiple donors. We found silencing in a 200 bp window around the transcription start site in polyclonal pools to be as effective as using wild-type Cas9 for identifying essential genes, but with much reduced cell numbers. Whole-genome screens to identify ARID1A-dependent dosage sensitivity revealed the PSMB2 gene, and enrichment of proteasome genes among the hits. This selective dependency was replicated with a proteasome inhibitor, indicating a targetable drug-gene interaction. Many more plausible targets in challenging cell models can be efficiently identified with our approach.
Keywords: human induced pluripotent stem cells, iPSCs, hiPSC, CRISPR, CRISPR-inhibition, CRISPRi, Genetic screening
Highlights
-
•
CRISPRi is as powerful as Cas9 for CRISPR screening in iPSCs
-
•
The silencing window of dCas9-KRAB-MeCP2 is can extend up to 1.4 kb away
-
•
Monoclones performed better, but polyclonal screens were sensitive enough to estimate big effects
-
•
Whole-genome CRISPRi screens identified silencing ARID1A makes iPSCs more prone to proteotoxic stress
The DNA break-induced stress upon Cas9 action limits the efficacy of whole-genome CRISPR screens in hiPSCs. Here, the authors established requirements for genome-wide CRISPR inhibition screens in hiPSC lines as a cost effective and efficient alternative to Cas9-based screens. By this method, they identified ARID1A gene interactions for stem cell growth including a targetable synthetic lethal interaction between the proteasome and ARID1A.
Introduction
Cell growth has to strike a balance between supporting development and survival across the lifespan against avoiding dysregulated proliferation. The genes required for this control have been mapped using genome-scale perturbation screens in different conditions and genetic backgrounds (Miles et al., 2016), and understanding them is important for restricting tumors as well as mitigating developmental disorders (Evers et al., 2016). There is an emerging consensus of the nature of gene essentiality from screens in many human cell lines, with a set of established core essential genes that are required for survival in most tested contexts, as well as a set of context-dependent vulnerabilities (Behan et al., 2019; Hart et al., 2015, 2017).
The CRISPR-Cas9 system has rapidly become the gold standard for pooled survival screens that collect this important information (Shalem et al., 2014; Wang et al., 2014). In brief, to knock out a gene, the Cas9 protein is directed to it by a guide RNA (gRNA), which results in a double-stranded DNA break that is ultimately repaired by error-prone pathways leading to small insertions and deletions that often disrupt the reading frame (Mali et al., 2013). These perturbations are relatively straightforward to parallelize, which enables efficient screening (Hanna and Doench, 2020). However, the DNA break generated by Cas9 can be toxic and trigger cell death, especially in stem cell contexts (Aguirre et al., 2016; Haapaniemi et al., 2018; Ihry et al., 2018; Peets et al., 2019; Rosenbluh et al., 2017). An alternative approach that does not suffer from this limitation is CRISPR inhibition (CRISPRi) (Gilbert et al., 2013; Qi et al., 2013), which employs a catalytically inactive Cas9 (dCas9) protein that is unable to cut DNA, and is usually fused to different effector domains to improve inhibition (Alerasool et al., 2020; Yeo et al., 2018). CRISPRi efficacy is known to be dependent on targeting the right transcript at the right distance to the transcription start site (TSS), but also to vary across contexts, which motivates evaluating and establishing these dependencies in the extensively used systems (Radzisheuskaya et al., 2016; Rosenbluh et al., 2017; Sanson et al., 2018; Yeo et al., 2018).
An important context to study disease mechanisms, drug targets, and development is stem cells, and in particular induced pluripotent stem cells (iPSCs) (Liu et al., 2020; Przybyla and Gilbert, 2022). The first genome-wide CRISPR-Cas screens in these cells have already shed light on mechanisms regulating embryonic stem cell survival and growth (Ihry et al., 2019; Mair et al., 2019; Peets et al., 2019). Cell types derived from stem cells via differentiation have been used to effectively and reversibly silence endogenous genes in cardiomyocytes (Mandegar et al., 2016), as well as to measure gene essentiality in neurons (Tian et al., 2019). Edited human iPSCs (hiPSCs) were also useful to identify early events of carcinogenesis before the accumulation of secondary mutations frequently observed in cancer lines (Wang et al., 2021). However, the potential of large-scale screening in stem cells is severely hampered by their cost, as the medium is expensive, and the cell death due to standard Cas9 action leads to requirements of very large numbers of cells. To unleash this potential, cost-permissive, large-scale, effective dropout CRISPRi screens have to be established in iPSCs, which requires understanding the efficacy of effector proteins and targeting constructs in this system. The current proof-of-principle studies are not yet informative enough for choosing the parameters for a screening campaign.
Here, we develop an effective CRISPRi screening approach in iPSCs, and use it to map mutation-specific and drug-sensitive genetic dependencies. First, we compare multiple fusion proteins, and identify dCas9-KRAB-MeCP2 as the most potent one. We then establish the characteristics of successful gRNAs that effectively repress their target using this fusion, identifying both the activity window as well as highlighting the importance of correct transcript selection. Next, we conduct whole-genome screens to determine the coverage requirements for an effective dropout screen in iPSCs, and compare these to standard CRISPR screens. Finally, we demonstrate the efficiency of CRISPRi screens to identify nocodazole-dependent gene essentiality, as well as genes that impact growth differently as a result of a mutation to the ARID1A cancer and developmental disorder gene.
Results
Effective gene silencing with CRISPRi depends on the fusion protein used and the choice of gRNAs. Therefore, we first explored the potency of dCas9 alternatives in hiPSCs, and identified the effect of gRNA features, such as proximity to different TSSs.
A rapid gene silencing efficacy test identifies the dCas9-KRAB-MeCP2 fusion protein as the most potent in hiPSCs
To identify the most potent dCas9 fusion construct, we compared three previously reported versions: dCas9-KRAB with piggyBAC delivery, KRAB-dCas9 with lentiviral delivery, and dCas9-KRAB-MeCP2 with piggyBAC delivery (Yeo et al., 2018). We produced polyclonal and monoclonal lines in iPSCs, and the human myeloid leukemia cell line K562 for each, and measured their efficacy in a dual fluorescence reporter system (Figures 1A, 1B, and S1). As expected, we observed reduced intensity of the targeted GFP gene in nearly all cases, with the most effective monoclonal lines outperforming the polyclonal ones (Figure 1C). In both K562 and iPSCs, the most successful clone was a monoclonal line with the dCas9-KRAB-MeCP2 construct, so we chose these lines for our screening applications, and retained the polyclones for reference as well.
dCas9-KRAB-MeCP2 silencing window can extend to 1.4 kb
After picking the fusion protein to work with, we characterized the determinants of successful gene repression with dCas9-KRAB-MeCP2 in iPSCs, starting with TSS distance. To determine the optimal target window, we designed a diverse gRNA library against various non-coding regions (including alternative TSSs) and coding sequences (Figure S2A).
We first targeted all NGG protospacer-adjacent motifs (PAMs) in 100 base pairs (bp) downstream of the TSS for 252 essential genes. On average, downregulation was most potent in the 20–40 bp downstream window (mean log2 fold change = −2.5), with a small but significant decrease of efficacy elsewhere (weakest mean log2 fold change = −2.2; Figure 2A). We next tiled all PAMs in a broader window of −200 to 300 bp around the TSS for 17 essential genes. We observed the strongest downregulation in the 0 to 100 bp window just downstream of the TSS, but more tempered effects further away (mean log2 fold change −2.05 vs. −1.81; Figure 2B).
Given the relative evenness of guide efficacy in these windows close to the TSS, we finally tested even more distal targets. We re-used a design that tiled coding sequences of 440 essential genes depleted in a genome-wide screen (Figure S2B) and observed significant depletion of at least 25% as far as 1.4 kb from the TSS, with effects further out indistinguishable from controls (Figure 2C). The decrease of efficacy was gradual, with an average log2 fold change for this library reducing from −1.2 within 100 bp of the TSS to −0.4 at 1.2–1.4 kb away. To validate the distance dependence of depletion phenotypes, we cloned six gRNAs targeting the essential ACTL6A gene at distances to its TSS ranging from 26 to 1,069 bp, infected a monoclonal dCas9-KRAB-MeCP2 iPSC line with them, and monitored growth under a live imaging system. We observed a decrease in growth rate that was proportional to TSS proximity (Figure 2D), confirming that the phenotypic impact can be modulated in the cell population by varying the distance of the targeting guide to the TSS.
Targeting the correct TSS is important for CRISPRi efficiency
As the TSS can be cell-type dependent, picking the correct one is required for designing efficient CRISPRi reagents. To measure the impact of TSS annotation, we picked genes with two TSSs separated by at least 200 bp from the tiling experiment, and calculated the difference in mean log2 fold change of the gRNAs targeting the two transcripts. We found that the CRISPRi phenotype change depended on the proximity of the two TSSs, with longer distances leading to larger differences in phenotype (Pearson’s R = 0.66, p < 10−15; Figure 2E).
Next we focused on 17 genes with alternative TSS annotation in iPSCs, with 13 of them showing minimum log2 fold change below −0.5 at either of the TSSs, and three of the genes having alternative TSSs at least 500 bp apart. Only one gene (CENPA, median log2 fold change 0.0 vs. −0.9), with more than 21 kb distance between transcripts, showed a differential repression effect, with the remaining 12 alternative TSSs efficiently targeted (Figures 2F and S2C). We then checked the TSS annotation of the 4,203 genes showing a depletion phenotype in any previous essentiality screens (Funk et al., 2022), and expressed in hiPSCs according to the FANTOM database (Lizio et al., 2015). Only 94 of the top TSSs according to FANTOM are targeted by gRNAs that are more distant than 1,400 bp, and, of these, only 10 are depleted in CRISPR (<−0.5 LFC) but not CRISPRi (>−1 LFC) (Figure S2D), which also include the two we tested separately above. We provide alternative gRNA designs for all of them in Table S1. Overall, TSS annotation is important for CRISPRi efficiency with the KRAB-MeCP2 construct, and consistent with observations from tiling experiments above, especially when the transcripts are separated by more than 1 kb.
CRISPRi screening in iPSCs efficiently identifies essential genes
To measure the performance of the dCas9-KRAB-MeCP2 construct in identifying essential genes on the genome scale, we conducted screens in hiPSC and K562 cell lines with the Dolcetto library (Sanson et al., 2018). We aimed for 200× coverage during infection, maintained 500× coverage throughout the remainder of the screen across technical replicates, and had mean coverage of 443 for sequencing libraries (Figures 3A and S3A; experimental procedures).
The whole-genome CRISPRi dropout screens successfully identified essential genes in both iPSCs (Figure 3) and K562 cells (Figure S3B). The screens were highly reproducible between replicates and different iPSC lines (Pearson’s R between gene average log2 fold changes >0.9; Figures 3B and 3C), and could successfully separate gold standard core essential and non-essential genes (Hart et al., 2014), with the area under the receiver-operator curve (AUC) above 0.9 in all cases (Figures 3D and 3E; Table S2). The performance at 50× coverage of a single replicate at day 18 post-infection was very close to that of a 200× screen combining four replicates (AUC = 0.977 vs. 0.978). There was little difference to another biological replicate (AUC = 0.978 vs. 0.980), as well as to combining information from multiple biological replicates (AUC = 0.980), or performing a single screen at 100× (AUC = 0.974). The separation of positive and negative controls increased with every passage, with the most performance gained when going from passage two at day 8 to passage three at day 14 (AUC = 0.957 vs. 0.977), and with little further improvement afterward (AUC = 0.980 at day 18). Altogether, 50× coverage screens in monoclonal lines measured at day 14 have near-optimal performance in iPSCs, but polyclonal lines and earlier timepoints can be used to trade quality for cost.
Producing monoclonal cell lines from a polyclonal pool requires at least 1 month of extra work and carries risks of additional positively selected mutations that can confound screening results. To test whether polyclonal cell populations suffice for high-quality screens, we conducted experiments in the parental polyclonal iPSC line with 70% silencing activity. We observed lower reproducibility of the biological replicates of polyclones (Pearson’s R of gene mean gRNA LFC = 0.91 vs. 0.99 in monoclones), as well as worse resolution at 50× coverage to separate essential and non-essential genes (median AUC = 0.97 vs. 0.98 in monoclones). However, this performance is still above the range reported in large-scale resources such as the Cancer Dependency Map (Behan et al., 2019), with all of the lines screened with wild-type Cas9 showing AUCs below 0.95. In addition, the log2 fold changes are highly correlated between mono- and polyclones (Pearson’s R = 0.92), Bayesian analysis approaches can overcome some of the reduction in the signal (Figures 3F and 3G), and 75% of genes identified by either of the screens are common to both (Figure 3H). Therefore, depending on the desired measurement precision, the cheaper and faster polyclonal screening will be sufficient for many purposes, and especially for identifying the strongest hits.
CRISPRi is as accurate as CRISPR in separating reference essential and non-essential genes
Next, we asked how the performance of the KRAB-MeCP2 CRISPRi construct compares with wild-type Cas9 screens. We previously conducted CRISPR-Cas9 screens in the same iPSC and K562 cell lines using a double gRNA library targeting 18,000 genes with 3 guides per gene (Peets et al., 2019). The CRISPRi screens compared favorably, with higher AUC values (0.98 for iPSC and 0.96 for K562) than CRISPR-Cas screens (0.95 for both lines; Figure 4A). The gene-level average gRNA log2 fold changes were moderately correlated between CRISPR and CRISPRi screens in the same cell line (Pearson’s R = 0.42), which is in line with concordances of both knockout and repression screens in different types of stem cells (R = 0.37–0.53; Figure S4A). Furthermore, of the essential genes identified in one screen, about a half were also hits in the other (1,143/2,292 for CRISPR and 1,143/2,159 for CRISPRi; Figure 4F). Overall, both systems can identify essential genes, but differences in cell line context, growth conditions, the library used, screen coverage, and other experimental parameters generate substantial variation in performance and hit lists both within and between various screen types. CRISPRi screens can be cheaper in stem cells due to the lack of double-strand break toxicity, as well as give reasonable performance using polyclonal lines.
CRISPR and CRISPRi screens employ different mechanisms of action to perturb the target, which can bias the identified hits. For example, targeting the X chromosome with wtCas9 in lines from male donors creates DNA breaks on only one chromosome and is less toxic to cells. Thus, guides targeting single-copy chromosomes are enriched in CRISPR screens but not CRISPRi screens (Figure 4B), while non-specific guides are more depleted (Figure 4C). On the other hand, the wider “blast radius” of CRISPRi can cause false-positive results (Rosenbluh et al., 2017). Indeed, when we compared nearby gene pairs with at most 1 kb distance between their TSSs, we observed a high correlation of signal in CRISPRi screens (R = 0.61) but not in CRISPR screens (Figure 4D), indicating a likely confounding due to targeting nearby TSSs.
The main drawback of CRISPR screens is double-strand break-induced cell death, which can be overcome by silencing TP53 and p53 pathway genes (Haapaniemi et al., 2018). We first tested whether targeting TP53 by CRISPRi in iPSCs also provides a growth advantage, and found it to confer a positive effect, but not as large as in CRISPR screens (log2 fold change upon TP53 perturbation 1.04 by CRISPR vs. 0.55 by CRISPRi; Figure S4B). Next, we confirmed the deleterious effect of double-strand breaks in CRISPR but not CRISPRi screens. Indeed, the average depletion of non-targeting control gRNAs was close to that of gRNAs targeting non-essential genes for CRISPRi screen (0.2 vs. 0.1; Figure 4G), but not CRISPR screen (0.68 vs. −1.02). This implies that targeting alone with a wtCas9 construct carries a phenotypic impact in hiPSCs (Aguirre et al., 2016; Haapaniemi et al., 2018). Finally, we validated that this effect is dose dependent, by comparing the growth of Cas9 and dCas9-KRAB-MeCP2 lines with guides that target zero, one, or multiple targets in the genome, and monitoring cell growth in a live imaging system. All three guides showed the same growth pattern in iPSC-dCas9-KRAB-MeCP2 lines, while the cutting at an additional number of targets caused increased cell death in iPSC-Cas9 cells (Figure 4D). Altogether, CRISPRi screens do not suffer from the dose-dependent double-strand break-induced cell death observed in CRISPR screens, but repressing the TP53 tumor suppressor still provides a moderate selective advantage.
CRISPRi identifies context-dependent essential genes
An important screening paradigm is identification of context-specific vulnerabilities for therapeutics and diagnostics. We therefore explored the utility of CRISPRi screens for mapping genome-scale drug-gene and allele-gene interactions. We first conducted pooled CRISPRi screens to identify genes that lead to nocodazole-dependent cell depletion (experimental procedures). Nocodazole is a well-established anti-mitotic agent that is less toxic in iPSCs. It arrests cells at G2/M phase by binding and destabilizing microtubules (Yiangou et al., 2019). The two genes most enriched in screens under nocodazole treatment compared with standard medium were TRIP13 and BRD9 (Figure 5A). TRIP13 is a gene implicated in cell-cycle control mechanisms, and its depletion is known to slow cell division, but to also allow cells to escape nocodazole-induced cell-cycle arrest (Ma and Poon, 2016; Marks et al., 2017). BRD9, however, has a TSS only 80 bp apart from TRIP13, suggesting that the BRD9 signal may be an unintended effect of the CRISPRi system and, indeed, BRD9 was only a hit using CRISPRi (LFC < −3) but not CRISPR screening (LFC = 0.16). Thus, the nocodazole-dependent signal from BRD9 is a likely false positive because of its TSS proximity to the known TRIP13 causal gene. Therefore, CRISPRi screening can identify gene-drug interactions, but the outputs need to be carefully analyzed to avoid modes of false positives not observed in CRISPR screens.
To identify mutation-specific gene effects, we screened for ARID1A-dependent vulnerabilities in iPSCs. The ARID1A gene produces a subunit of the ATP-dependent histone deacetylase SWI/SNF complex that has been implicated in developmental disorders, as well as 20% of all cancers (Pagliaroli and Trizzino, 2021) therefore potentially an important synthetic lethal gene dependency target for iPSCs. We produced isogenic iPSC lines by targeting the ARID1A coding region with transiently expressed Cas9, and picking one clonal line with a 7-nucleotide heterozygous deletion (c. 2389–2395: TCCAGCAGC>TC, p. 667–669: SSS/X), and one control line with wild-type sequence. The heterozygous mutation decreased ARID1A expression by 60% and resulted in slower growth compared with the wild-type line (Figure 5B). To prepare the cells for CRISPRi screening, we transfected the two lines with the dCas9-KRAB-MeCP2 construct and selected monoclonal cell lines with at least 65% silencing activity in the reporter assay. We proceeded with genome-wide screens in mutant and wild-type backgrounds using the Dolcetto library as described above. The depletion signals of the two screens were less correlated compared with replicates in wild-type cells (Pearson’s R of BAGEL Bayes factors = 0.87 vs. 0.94; Figure 5C), suggesting that the ARID1A mutation altered the gene essentiality profile.
We next aimed to understand the differentially depletion-sensitive genes. Tumor suppressor gene NF2 and chromatin remodeler gene TAF6L had the strongest increase of statistical signal in the mutant line (experimental procedures), while genes implicated in RNA and protein turnover (EXOSC7 and SEL1L) and protein trafficking (AP1G1) had the most decrease. SEL1L encodes the adaptor subunit of ERAD ubiquitin ligase, which extracts misfolded proteins from the endoplasmic reticulum into the cytosol to be degraded by the proteasome (Hwang and Qi, 2018). Consistent with the role of protein degradation, inhibition of the proteasome 20S subunit beta 2 (PSMB2) gene, a potential drug target, resulted in more cell death in the ARID1A+/− line (Figure 5B). To confirm this differential growth effect of proteasome inhibition on ARID1A wild-type and mutant lines, we applied physiological concentrations of oprozomib, a second-generation proteasome inhibitor in trial for the treatment of hematological cancers (Sherman and Li, 2020), and measured the difference in growth rate. We found that ARID1A mutant line decreased in cell confluence more than wild-type cells upon oprozomib treatment when compared with the no drug control (Figure 5D). Our results thus confirm the sensitivity of ARID1A mutant cancer lines to proteotoxic stress (Tomihara et al., 2021), which could be therapeutically exploited.
More globally, gene set enrichment analysis of all the genes ranked by the difference between the wild-type and mutant line screen depletion identified the proteasome core complex among the top 20 enriched pathways (Figure 5E), as well as other pathways related to ARID1A function, like centromere complex assembly and kinetochore organization, confirming previous results (Caumanns et al., 2018; Mathur, 2018). Furthermore, a String database (Szklarczyk et al., 2021) analysis with 200 most depleted genes identified three clusters: gene expression regulation, cell-cycle regulation, and cholesterol biosynthesis regulation (Figure S5) around ARID1A and two other members of SWI/SNF family (SS18 and SMARCE1). We also tested whether essentiality changes upon ARID1A dose reduction mimic those in cancer cell lines upon damaging mutations to the ARID1A gene, but found no strong signal (Note S1).
Context-dependent vulnerabilities replicate with different reagents and assays
We confirmed the genotype- and small-molecule-dependent fitness effects with alternative reagents and assays. First, to validate that the reduced dose of ARID1A, rather than any other clonal mutation, leads to growth rate reduction in the ARID1A mutant line, we tested six gRNAs targeting 17, 59, 894, 987, 1,160, and 1,314 nt downstream of the TSS of ARID1A in wild-type iPSCs (experimental procedures). Two of the gRNAs, targeting 17 and 1,314 nt downstream of TSS, caused substantial growth retardation in the first 36 h after transfection (Figure 6A).
Second, we confirmed the impact of silencing EZH2, coding for the catalytic domain of histone methyltransferase polycomb repressive complex 2 (Figure 5B), which was more essential in the wild-type screen compared with the one in the ARID1A mutant line. We transfected cells with either EZH2 targeting or non-targeting gRNAs and, after 9 days in culture, observed substantial morphology changes upon targeting compared with non-targeting guides in ARID1A+/+ cells and EZH2-silenced ARID1A+/− cells (Figure S6A). Furthermore, 32% of EZH2-silenced ARID1A+/+ cells were negative for the stem cell marker hTRA-1-60 (Figure 6B), indicating likely differentiation from stem cell state toward a less propagating lineage with different morphology and expression profile, and implicating ARID1A as a causal intermediate factor in this process.
Third, we validated hits from the whole-genome screens. We compared the signal between CRISPRi screens in ARID1A+/+ and ARID1A+/− lines and picked 1,657 genes with large differences (experimental procedures) to re-measure in a follow-up experiment. Screens in the wild-type and mutant lines using a gRNA library against these genes were consistent with the original ones both in absolute terms (Pearson’s R of gene LFCs > 0.84; Figures 6C and S6B), as well as for difference in the mutant line (Pearson’s R of rank changes 0.75; Figure 6D).
Finally, to confirm that sensitivity to proteasome inhibition is due to ARID1A gene targeting, we treated the ARID1A wild-type lines with the two strongly growth-retarding ARID1A-targeting gRNAs, one weak gRNA, a negative control gRNA, and no gRNA, as well as with oprozomib (1 μg/mL) for 1 day and compared growth with non-treated controls, and the ARID1A+/− line. The vulnerability of hiPSCs to oprozomib treatment was dependent on ARID1A targeting, with gRNAs leading to stronger growth effect in hiPSCs also giving rise to stronger phenotypes upon proteasome inhibition, resembling the growth of the mutant line (Figure 6E). Thus, we have demonstrated that an important developmental disorder and cancer gene can potentially be therapeutically targeted with a small molecule in a dose-dependent manner.
Discussion
We presented the first systematic investigation of genome-wide CRISPRi screens in hiPSCs and demonstrated that CRISPRi is an efficient and safe alternative to standard CRISPR-Cas-based screening. We identified dCas9-KRAB-MeCP2 as the most potent fusion protein for target silencing in hiPSCs, with a range that can extend to 1.4 kb into the coding region. Whole-genome CRISPRi screens using this construct performed as well as different novel and previously published screens with standard Cas9. We identified actionable drug sensitivity for ARID1A mutant iPSCs, and other interactions illuminating the role of ARID1A for stem cell growth and differentiation.
The best silencing modality is not easy to choose for a new experiment. For example, dCas9-KRAB is effective when integrated into the AAVS1 locus in hiPSCs (Mandegar et al., 2016), but replicating this setup is cumbersome for screening in multiple lines. The dCas9-KRAB-MeCP2 performed even better for silencing (Yeo et al., 2018), but varied in efficiency across cell types. By testing three constructs with different delivery methods and both monoclonal and polyclonal versions, we found that monoclonal lines expressing the dCas9-KRAB-MeCP2 fusion protein showed both best silencing activity with the reporter construct and best performance in genome-wide screens.
Another important consideration in practice is the cost required for screening at scale. Both standard CRISPR and CRISPRi screens can identify essential genes in an experiment (Gilbert et al., 2014; Horlbeck et al., 2016; Sanson et al., 2018), and while CRISPR has previously been argued to be more sensitive (Rosenbluh et al., 2017; Sanson et al., 2018), CRISPRi outperformed CRISPR in both hiPSC and K562 cells in our hands, at similar screen coverages and screen durations. In particular, we observed that a 14-day duplicate screen with 50× coverage at infection and 200× during passaging gives near-optimal results, greatly reducing the cost per screen compared with more standard larger designs. Efficacy, resolution, budget, and labor for a screen can be optimized by choosing mono- or polyclonal lines, screen coverage, and duration.
The dCas9 protein has to be targeted near the TSS for efficient screening (Gilbert et al., 2013), but the blast radius varies for different fusion proteins. The best window for dCas9-KRAB activity has been established to be 0 to 100 bp after the TSS (Radzisheuskaya et al., 2016; Rosenbluh et al., 2017; Sanson et al., 2018; Yeo et al., 2018). However, these reports are specific to the KRAB fusion and explored only a few target sites. We demonstrated that the target range of dCas9-KRAB-MeCP2 in iPSCs is larger, with impacts on gene expression as far as 1.4 kb after the TSS, and with the efficiency gradually decreasing with distance. We propose that this understanding could be used as a way to modulate the extent of downregulation, as has been previously done with mismatched gRNAs (Jost et al., 2020). This fine-grained control would be useful to study haploinsufficiency of target genes, and other questions that require a more precise range of the gene dose. The large impact range of dCas9-KRAB-MeCP2 also decreases the false-negative rate due to alternative TSSs for different cell types. We showed that false-negative results rarely occur for the majority of genes for which the alternative is less than 1 kb away.
CRISPR screens have already yielded gene-gene interactions (Horlbeck et al., 2018; Kim et al., 2022). We screened for interactions with the ARID1A gene, implicated in both developmental disorders (OMIM: 135900) as well as cancer (Pagliaroli and Trizzino, 2021), suggesting growth control to be an important aspect of its function. Suppressing the NF2 gene in the ARID1A mutant background gave iPSCs a growth advantage and, although direct interaction of ARID1A and NF2 is not known, both proteins regulate pathways inhibiting oncogenic YAP/TAZ genes, and their double mutation causes hepatocellular carcinoma in mice (Chang et al., 2018; Patel et al., 2017). Two other hits, TAF6L and EZH2, are members of the SAGA and PRC2 chromatin-modifying complexes, interacting with SWI/SNF in cell fate determination (Pagliaroli and Trizzino, 2021; Seruggia et al., 2019). Further investigation of the EZH2 effect with an arrayed screen demonstrated the growth advantage in the ARID1A+/− background was due to inhibition of cell differentiation, which further illuminates ARID1A function in cell fate determination.
Pluripotent stem cells of different provenance and capacity are poised to expand our understanding and ability to engineer disease mutations, cell types, entire organs, and courses of development (Hanna et al., 2002; Liu et al., 2020). Precise methods for genome engineering to enable control in these systems is a crucial enabling technology for this progress. Detection of context-specific effects will be key both for accurately targeting only the diseased state, as well as to chart the inherent heterogeneity that needs to be accounted for.
Experimental procedures
Resource availability
Corresponding authors
Further inquiries and requests for data and resources should be directed to corresponding authors, Sunay Usluer (sunay.usluer@sanger.ac.uk) and Leopold Parts (leopold.parts@sanger.ac.uk).
Materials availability
Libraries and plasmids uniquely produced through this study will be made available upon request and will require a material transfer agreement.
Cell culture
K562-Cas9 and K562-Cas9-GFP cells were cultured in RPMI supplemented with 10% FCS, 2 mM L-glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin (will be referred to as supplemented RPMI medium hereafter). Cells were passaged 1:20 every 4 days. Wild-type hiPSC (fiaj-1 and kolf2) lines were obtained from the HipSci platform, where its karyotype and phenotype were evaluated to be normal. hiPSC lines were cultured on Vitronectin-XF (STEMCELL Technologies, 07180)-coated plates and in mTeSR-Plus Media (STEMCELL Technologies, 100–0276). For maintenance, they were passaged every 4 days as clumps using ReLeSR (STEMCELL Technologies, 05872) according to the producer’s protocol. All cell lines were cultured at 37°C, 5% CO2.
Cloning stable dCas9 fusion protein-expressing lines
Three different dCas9 fusion proteins were tested for dCas9 activity in both K562 and hiPSC lines: pLX_311-KRAB-dCas9 (Addgene 96918), pB-CAGGS-dCas9-KRAB-MeCP2 (Addgene, 110824), and pPB-dCas9-KRAB (kindly provided by Dr. Qianxin Wu). The first is a lentiviral delivery vector, whereas the latter two rely on the piggyBac transposase for genome integration. We produced hiPSC and K562 lines stably expressing all three fusion proteins by transducing/transfecting them with the above vectors (supplemental experimental procedures).
Producing monoclonal hiPSC lines
After 10 days of blasticidin selection, polyclonal cell populations were collected with accutase and 1,000 cells seeded into 6-cm dishes precoated with Synthemax II-SC Substrate (Corning, 3535) at a concentration of 5 μg/cm2 in mTeSR-E8 medium supplemented with 10× CloneR (Stem Cells, 05888). The medium was changed every day with blasticidin-supplemented mTeSR+Plus medium (10 μg/m). After 10 days, visible colonies were manually transferred into 12-well plates precoated with Vitronectin XF. Surviving monoclonal cell lines were expanded and used in later analyses.
Producing monoclonal K562 lines
After 10 days of blasticidin selection, monoclonal lines were produced by serial dilution as follows: polyclonal cells were counted and serial diluted to an estimated 50 cells in 20 mL K562 medium and 200 μL of suspension was distributed into each well of a 96-well plate. In the first iteration, there was cell growth in 6 wells. Cells from these 6 wells were separately collected, diluted, and redistributed into 6 different 96-well plates, and one well from each plate was picked as monoclonal lines.
Testing CRISPRi efficacy
Monoclonal and polyclonal hiPSC-dCas9 lines were transfected with pCS2-iREP-GFP-PGK-BFP-U6-gRNA-irep or pCS2-iREP-GFP-PGK-BFP-U6-gRNA-mock vectors (cloning details are in supplemental experimental procedures) as replicates in 12-well plates using TransIT-LT1 Transfection Reagent (Mirus Bio, MIR2300) according to the manufacturer’s instructions. Similarly, monoclonal and polyclonal K562-dCas9 lines were transfected with pCS2-iREP-GFP-PGK-BFP-U6-gRNA-irep or pCS2-iREP-GFP-PGK-BFP-U6-gRNA-mock vectors in replicate with Lipofectamine LTX reagent, with non-fluorescent plasmid as negative control. Three days after transfection, hiPSCs were harvested with accutase and K562 cells were collected by centrifugation. All cells were washed and resuspended in PBS+FBS (2%). Harvested cells were analyzed in a CytoFLEX flow cytometer using FlowJo analysis software (Beckman Coulter). CRISPRi efficacy was quantified as percent activity – the percent decrease in median GFP level in BFP-positive cells.
Generating the CRISPRi tiling library
Design
To determine the optimal target window of dCas9-KRAB-MeCP2 in iPSCs, three sub-libraries were designed (Figure S2A). First, the TSS tiling library placed guides relative to the TSSs in hiPSCs, as identified from the CAGE annotations in the Fantom database (Lizio et al., 2015) according to the GRCh38 reference genome. Within the TSS tiling library, the “TSS Tiling, Extra” library tiled from 0 to +100 bp of TSS site for the single transcript of 882 genes essential in hiPSCs (Figure 2A); the “TSS Tiling, Different” library tiled from −200 to +300 bp of the TSS site for the top 2 transcripts of 20 essential genes for which the TSS annotation in hiPSCs did not match the canonical one (Figure 2B); and the “TSS Tiling, Other” library tiled the top 2 transcripts of 20 essential genes. Second, the Gene tiling library targeted guides to all PAMs in coding sequence of a single transcript of 451 Hart essential genes, and 36 non-essential genes (Figure 2C). Finally, 200 non-targeting guides were used as control (Table S1).
The designed library was produced as complex oligonucleotide pools (Genescript) and cloned into lentiviral gRNA expression vector pKLV2-U6gRNA5(BbsI)-ccdb-PGKpuroBFP-W (AddGene: 67974) (Tzelepis et al., 2016) (supplemental experimental procedures).
gRNAs for iPSC-specific alternative TSSs for 10 genes were designed using the CRISPick tool of the Broad Institute (Doench et al., 2016; Sanson et al., 2018).
CRISPRi screening
The Dolcetto library (Sanson et al., 2018) was acquired from Addgene (no. 1000000114) as two plasmid pools, each with a half-library (sets A and B) in the XPR_500 backbone. Only the Dolcetto set A plasmid pool was used in this study. Plasmid pools were re-amplified, packed into lentivirus, and titrated as detailed in supplemental experimental procedures.
hiPSCs
Cells (6 × 107) were mixed with the lentiviral Dolcetto library to achieve an estimated 0.3 MOI and seeded on 2× vitronectin-coated 5-layer flasks at 36K cells/cm2 density in mTeSR-Plus medium with Rock inhibitor (Ri) (10 μM, Y-27632, STEMCELL Technologies, 72304). The next day, the medium was changed to remove Ri. On day 3 post infection, puromycin selection was started at 0.5 μg/mL, and continued through the screen duration (17–22 days). Cells were passaged at 90% confluency and seeded back at 16K cells/cm2 density for the rest of the screen. The remaining cell pellets were aliquoted and frozen for genomic DNA extraction.
K562 cells
Cells (75 × 106) were resuspended in supplemented RPMI+ polybrene (8 μg/mL) medium and mixed with the Dolcetto library virus aiming for a MOI of 0.3. The cell + virus suspension was centrifuged at 1,000 rcf for 30 min. The cells were then resuspended in the same medium and plated into two T150cm2 plates for a final concentration of 0.45 × 106 cells/mL. The infected cells were selected with puromycin (2 μg/mL) starting from day 1 after infection until day 8. After complete selection, the cells in culture were passaged every 3 or 4 days, and plated back at 0.1 × 106 cells/mL density. Aliquoted pellets of at least 40 × 106 cells were frozen for genomic DNA extraction.
Genomic DNA isolation and sequencing library preparation were conducted as reported previously (Peets et al., 2019) and as in supplemental experimental procedures.
Data analysis
gRNA sequences were counted from fasta files and aligned to corresponding library maps without allowing any mismatches. Uniformity of plasmid libraries was determined using the Gini index as implemented by Wang et al. (2019). Average sequencing coverage of all libraries was around 400×. When all technical replicates were combined together, a single screen had around 2,000× coverage for final time points. Raw counts in each sequencing sample were normalized and transformed to log2 scale according to the formula: log2(((reads per guide/sum of all read counts) × 106) + 1). Log2 fold change per guide was calculated by subtracting log2 transformed and normalized read counts of final time points from plasmid library sequences. To determine essential and non-essential genes, Bagel v.2 (Kim and Hart, 2021) was used with default parameters. To find differentially essential genes between ARID1A+/+ and ARID1A +/− lines, we calculated the mean and standard deviation of the difference across all genes, and assigned the genes with values of at least two standard deviations away from the mean to be differentially essential.
Author contributions
S.U. and L.P. conceptualized and initiated the study with the help of L.C. S.U. performed experiments with the help of Y.Z., K.U., and C.D. S.U. and P.H. analyzed the data with the help of J.S., G.N., and O.E.G. K.A. designed the TSS tiling library. B.N. created isogenic hiPSC lines with the supervision of S.S.G. and O.M.D. L.P. supervised the project. S.U. and L.P. wrote the manuscript with input from all authors.
Acknowledgments
S.U. was supported by the Janet Thornton Fellowship. P.H., L.C., Y.Z., K.U., C.D., J.S., G.N., O.E.G., S.G., B.N., O.M.D., and L.P. were supported by Wellcome (grant nos. 108413/A/15/D and 220540/Z/20/A, “Wellcome Sanger Institute Quinquennial Review 2021–2026”). K.A. was supported by the Estonian Research Council (PSG415) and the Archimedes Foundation (EXCITE TK148).
The authors acknowledge the Wellcome Trust Sanger Institute as the source of (fiaj-1 and kolf-2) human induced pluripotent cell lines, which were generated under the Human Induced Pluripotent Stem Cell Initiative funded by a grant from the Wellcome Trust and the Medical Research Council, supported by the Wellcome Trust (WT098051) and the NIHR/Wellcome Trust Clinical Research Facility, and also acknowledge the Life Science Technologies Corporation as set out at Schedule 1.
Conflict of interests
L.C. and O.M.D. receive remuneration and stock options from bit.bio.
Published: April 6, 2023
Footnotes
Supplemental information can be found online at https://doi.org/10.1016/j.stemcr.2023.03.008.
Contributor Information
Sunay Usluer, Email: sunay.usluer@sanger.ac.uk.
Leopold Parts, Email: leopold.parts@sanger.ac.uk.
Supplemental information
Data and code availability
Data and analyses for this study are available from the following git hub repository: https://github.com/sunayusluer/CRISPRi-Analysis.
References
- Aguirre A.J., Meyers R.M., Weir B.A., Vazquez F., Zhang C.-Z., Ben-David U., Cook A., Ha G., Harrington W.F., Doshi M.B., et al. Genomic copy number dictates a gene-independent cell response to CRISPR/Cas9 targeting. Cancer Discov. 2016;6:914–929. doi: 10.1158/2159-8290.CD-16-0154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Alerasool N., Segal D., Lee H., Taipale M. An efficient KRAB domain for CRISPRi applications in human cells. Nat. Methods. 2020;17:1093–1096. doi: 10.1038/s41592-020-0966-x. [DOI] [PubMed] [Google Scholar]
- Behan F.M., Iorio F., Picco G., Gonçalves E., Beaver C.M., Migliardi G., Santos R., Rao Y., Sassi F., Pinnelli M., et al. Prioritization of cancer therapeutic targets using CRISPR–Cas9 screens. Nature. 2019;568:511–516. doi: 10.1038/s41586-019-1103-9. [DOI] [PubMed] [Google Scholar]
- Caumanns J.J., Wisman G.B.A., Berns K., van der Zee A.G.J., de Jong S. ARID1A mutant ovarian clear cell carcinoma: a clear target for synthetic lethal strategies. Biochim. Biophys. Acta. Rev. Cancer. 2018;1870:176–184. doi: 10.1016/j.bbcan.2018.07.005. [DOI] [PubMed] [Google Scholar]
- Chang L., Azzolin L., Di Biagio D., Zanconato F., Battilana G., Lucon Xiccato R., Aragona M., Giulitti S., Panciera T., Gandin A., et al. The SWI/SNF complex is a mechanoregulated inhibitor of YAP and TAZ. Nature. 2018;563:265–269. doi: 10.1038/s41586-018-0658-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Doench J.G., Fusi N., Sullender M., Hegde M., Vaimberg E.W., Donovan K.F., Smith I., Tothova Z., Wilen C., Orchard R., et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nat. Biotechnol. 2016;34:184–191. doi: 10.1038/nbt.3437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Evers B., Jastrzebski K., Heijmans J.P.M., Grernrum W., Beijersbergen R.L., Bernards R. CRISPR knockout screening outperforms shRNA and CRISPRi in identifying essential genes. Nat. Biotechnol. 2016;34:631–633. doi: 10.1038/nbt.3536. [DOI] [PubMed] [Google Scholar]
- Funk L., Su K.-C., Ly J., Feldman D., Singh A., Moodie B., Blainey P.C., Cheeseman I.M. The phenotypic landscape of essential human genes. Cell. 2022;185:4634–4653.e22. doi: 10.1016/j.cell.2022.10.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gilbert L.A., Larson M.H., Morsut L., Liu Z., Brar G.A., Torres S.E., Stern-Ginossar N., Brandman O., Whitehead E.H., Doudna J.A., et al. CRISPR-mediated modular RNA-guided regulation of transcription in eukaryotes. Cell. 2013;154:442–451. doi: 10.1016/j.cell.2013.06.044. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gilbert L.A., Horlbeck M.A., Adamson B., Villalta J.E., Chen Y., Whitehead E.H., Guimaraes C., Panning B., Ploegh H.L., Bassik M.C., et al. Genome-scale CRISPR-mediated control of gene repression and activation. Cell. 2014;159:647–661. doi: 10.1016/j.cell.2014.09.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Haapaniemi E., Botla S., Persson J., Schmierer B., Taipale J. CRISPR–Cas9 genome editing induces a p53-mediated DNA damage response. Nat. Med. 2018;24:927–930. doi: 10.1038/s41591-018-0049-z. [DOI] [PubMed] [Google Scholar]
- Hanna R.E., Doench J.G. Design and analysis of CRISPR-Cas experiments. Nat. Biotechnol. 2020;38:813–823. doi: 10.1038/s41587-020-0490-7. [DOI] [PubMed] [Google Scholar]
- Hanna L.A., Foreman R.K., Tarasenko I.A., Kessler D.S., Labosky P.A. Requirement for Foxd3 in maintaining pluripotent cells of the early mouse embryo. Genes Dev. 2002;16:2650–2661. doi: 10.1101/gad.1020502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hart T., Brown K.R., Sircoulomb F., Rottapel R., Moffat J. Measuring error rates in genomic perturbation screens: gold standards for human functional genomics. Mol. Syst. Biol. 2014;10:733. doi: 10.15252/msb.20145216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hart T., Chandrashekhar M., Aregger M., Steinhart Z., Brown K.R., MacLeod G., Mis M., Zimmermann M., Fradet-Turcotte A., Sun S., et al. High-resolution CRISPR screens reveal fitness genes and genotype-specific cancer liabilities. Cell. 2015;163:1515–1526. doi: 10.1016/j.cell.2015.11.015. [DOI] [PubMed] [Google Scholar]
- Hart T., Tong A.H.Y., Chan K., Van Leeuwen J., Seetharaman A., Aregger M., Chandrashekhar M., Hustedt N., Seth S., Noonan A., et al. Evaluation and design of genome-wide CRISPR/SpCas9 knockout screens. G3. 2017;7:2719–2727. doi: 10.1534/g3.117.041277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Horlbeck M.A., Gilbert L.A., Villalta J.E., Adamson B., Pak R.A., Chen Y., Fields A.P., Park C.Y., Corn J.E., Kampmann M., Weissman J.S. Compact and highly active next-generation libraries for CRISPR-mediated gene repression and activation. Elife. 2016;5 doi: 10.7554/eLife.19760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Horlbeck M.A., Xu A., Wang M., Bennett N.K., Park C.Y., Bogdanoff D., Adamson B., Chow E.D., Kampmann M., Peterson T.R., et al. Mapping the genetic landscape of human cells. Cell. 2018;174:953–967.e22. doi: 10.1016/j.cell.2018.06.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hwang J., Qi L. Quality control in the endoplasmic reticulum: crosstalk between ERAD and UPR pathways. Trends Biochem. Sci. 2018;43:593–605. doi: 10.1016/j.tibs.2018.06.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ihry R.J., Worringer K.A., Salick M.R., Frias E., Ho D., Theriault K., Kommineni S., Chen J., Sondey M., Ye C., et al. p53 inhibits CRISPR–Cas9 engineering in human pluripotent stem cells. Nat. Med. 2018;24:939–946. doi: 10.1038/s41591-018-0050-6. [DOI] [PubMed] [Google Scholar]
- Ihry R.J., Salick M.R., Ho D.J., Sondey M., Kommineni S., Paula S., Raymond J., Henry B., Frias E., Wang Q., et al. Genome-scale CRISPR screens identify human pluripotency-specific genes. Cell Rep. 2019;27:616–630.e6. doi: 10.1016/j.celrep.2019.03.043. [DOI] [PubMed] [Google Scholar]
- Jost M., Santos D.A., Saunders R.A., Horlbeck M.A., Hawkins J.S., Scaria S.M., Norman T.M., Hussmann J.A., Liem C.R., Gross C.A., Weissman J.S. Titrating gene expression using libraries of systematically attenuated CRISPR guide RNAs. Nat. Biotechnol. 2020;38:355–364. doi: 10.1038/s41587-019-0387-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kim E., Hart T. Improved analysis of CRISPR fitness screens and reduced off-target effects with the BAGEL2 gene essentiality classifier. Genome Med. 2021;13:2. doi: 10.1186/s13073-020-00809-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kim E., Novak L.C., Lin C., Colic M., Bertolet L.L., Gheorghe V., Bristow C.A., Hart T. Dynamic rewiring of biological activity across genotype and lineage revealed by context-dependent functional interactions. Genome Biol. 2022;23:140. doi: 10.1186/s13059-022-02712-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Liu G., David B.T., Trawczynski M., Fessler R.G. Advances in pluripotent stem cells: history, mechanisms, technologies, and applications. Stem Cell Rev. Rep. 2020;16:3–32. doi: 10.1007/s12015-019-09935-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lizio M., Harshbarger J., Shimoji H., Severin J., Kasukawa T., Sahin S., Abugessaisa I., Fukuda S., Hori F., Ishikawa-Kato S., et al. Gateways to the FANTOM5 promoter level mammalian expression atlas. Genome Biol. 2015;16:22. doi: 10.1186/s13059-014-0560-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ma H.T., Poon R.Y.C. TRIP13 regulates both the activation and inactivation of the spindle-assembly checkpoint. Cell Rep. 2016;14:1086–1099. doi: 10.1016/j.celrep.2016.01.001. [DOI] [PubMed] [Google Scholar]
- Mair B., Tomic J., Masud S.N., Tonge P., Weiss A., Usaj M., Tong A.H.Y., Kwan J.J., Brown K.R., Titus E., et al. Essential gene profiles for human pluripotent stem cells identify uncharacterized genes and substrate dependencies. Cell Rep. 2019;27:599–615.e12. doi: 10.1016/j.celrep.2019.02.041. [DOI] [PubMed] [Google Scholar]
- Mali P., Yang L., Esvelt K.M., Aach J., Guell M., DiCarlo J.E., Norville J.E., Church G.M. RNA-guided human genome engineering via Cas9. Science. 2013;339:823–826. doi: 10.1126/science.1232033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mandegar M.A., Huebsch N., Frolov E.B., Shin E., Truong A., Olvera M.P., Chan A.H., Miyaoka Y., Holmes K., Spencer C.I., et al. CRISPR interference efficiently induces specific and reversible gene silencing in human iPSCs. Cell Stem Cell. 2016;18:541–553. doi: 10.1016/j.stem.2016.01.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Marks D.H., Thomas R., Chin Y., Shah R., Khoo C., Benezra R. Mad2 overexpression uncovers a critical role for TRIP13 in mitotic exit. Cell Rep. 2017;19:1832–1845. doi: 10.1016/j.celrep.2017.05.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mathur R. ARID1A loss in cancer: towards a mechanistic understanding. Pharmacol. Ther. 2018;190:15–23. doi: 10.1016/j.pharmthera.2018.05.001. [DOI] [PubMed] [Google Scholar]
- Miles L.A., Garippa R.J., Poirier J.T. Design, execution, and analysis of pooled in vitro CRISPR/Cas9 screens. FEBS J. 2016;283:3170–3180. doi: 10.1111/febs.13770. [DOI] [PubMed] [Google Scholar]
- Pagliaroli L., Trizzino M. The evolutionary conserved SWI/SNF subunits ARID1A and ARID1B are key modulators of pluripotency and cell-fate determination. Front. Cell Dev. Biol. 2021;9 doi: 10.3389/fcell.2021.643361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Patel S.H., Camargo F.D., Yimlamai D. Hippo signaling in the liver regulates organ size, cell fate, and carcinogenesis. Gastroenterology. 2017;152:533–545. doi: 10.1053/j.gastro.2016.10.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Peets E.M., Crepaldi L., Zhou Y., Allen F., Elmentaite R., Noell G., Turner G., Iyer V., Parts L. Minimized double guide RNA libraries enable scale-limited CRISPR/Cas9 screens. bioRxiv. 2019 doi: 10.1101/859652. Preprint at. [DOI] [Google Scholar]
- Przybyla L., Gilbert L.A. A new era in functional genomics screens. Nat. Rev. Genet. 2022;23:89–103. doi: 10.1038/s41576-021-00409-w. [DOI] [PubMed] [Google Scholar]
- Qi L.S., Larson M.H., Gilbert L.A., Doudna J.A., Weissman J.S., Arkin A.P., Lim W.A. Repurposing CRISPR as an RNA-guided platform for sequence-specific control of gene expression. Cell. 2013;152:1173–1183. doi: 10.1016/j.cell.2013.02.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Radzisheuskaya A., Shlyueva D., Müller I., Helin K. Optimizing sgRNA position markedly improves the efficiency of CRISPR/dCas9-mediated transcriptional repression. Nucleic Acids Res. 2016;44:e141. doi: 10.1093/nar/gkw583. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rosenbluh J., Xu H., Harrington W., Gill S., Wang X., Vazquez F., Root D.E., Tsherniak A., Hahn W.C. Complementary information derived from CRISPR Cas9 mediated gene deletion and suppression. Nat. Commun. 2017;8:15403–15408. doi: 10.1038/ncomms15403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sanson K.R., Hanna R.E., Hegde M., Donovan K.F., Strand C., Sullender M.E., Vaimberg E.W., Goodale A., Root D.E., Piccioni F., Doench J.G. Optimized libraries for CRISPR-Cas9 genetic screens with multiple modalities. Nat. Commun. 2018;9:5416. doi: 10.1038/s41467-018-07901-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Seruggia D., Oti M., Tripathi P., Canver M.C., LeBlanc L., Di Giammartino D.C., Bullen M.J., Nefzger C.M., Sun Y.B.Y., Farouni R., et al. TAF5L and TAF6L maintain self-renewal of embryonic stem cells via the MYC regulatory network. Mol. Cell. 2019;74:1148–1163.e7. doi: 10.1016/j.molcel.2019.03.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shalem O., Sanjana N.E., Hartenian E., Shi X., Scott D.A., Mikkelson T., Heckl D., Ebert B.L., Root D.E., Doench J.G., Zhang F. Genome-scale CRISPR-Cas9 knockout screening in human cells. Science. 2014;343:84–87. doi: 10.1126/science.1247005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sherman D.J., Li J. Proteasome inhibitors: harnessing proteostasis to combat disease. Molecules. 2020;25 doi: 10.3390/molecules25030671. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Szklarczyk D., Gable A.L., Nastou K.C., Lyon D., Kirsch R., Pyysalo S., Doncheva N.T., Legeay M., Fang T., Bork P., et al. The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Res. 2021;49:D605–D612. doi: 10.1093/nar/gkaa1074. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tian R., Gachechiladze M.A., Ludwig C.H., Laurie M.T., Hong J.Y., Nathaniel D., Prabhu A.V., Fernandopulle M.S., Patel R., Abshari M., et al. CRISPR interference-based platform for multimodal genetic screens in human iPSC-derived neurons. Neuron. 2019;104:239–255.e12. doi: 10.1016/j.neuron.2019.07.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tomihara H., Carbone F., Perelli L., Huang J.K., Soeung M., Rose J.L., Robinson F.S., Lissanu Deribe Y., Feng N., Takeda M., et al. Loss of ARID1A promotes epithelial–mesenchymal transition and sensitizes pancreatic tumors to proteotoxic stress. Cancer Res. 2021;81:332–343. doi: 10.1158/0008-5472.CAN-19-3922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tzelepis K., Koike-Yusa H., De Braekeleer E., Li Y., Metzakopian E., Dovey O.M., Mupo A., Grinkevich V., Li M., Mazan M., et al. A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia. Cell Rep. 2016;17:1193–1205. doi: 10.1016/j.celrep.2016.09.079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang B., Wang M., Zhang W., Xiao T., Chen C.-H., Wu A., Wu F., Traugh N., Wang X., Li Z., et al. Integrative analysis of pooled CRISPR genetic screens using MAGeCKFlute. Nat. Protoc. 2019;14:756–780. doi: 10.1038/s41596-018-0113-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang T., Wei J.J., Sabatini D.M., Lander E.S. Genetic screens in human cells using the CRISPR-Cas9 system. Science. 2014;343:80–84. doi: 10.1126/science.1246981. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang T., Pine A.R., Kotini A.G., Yuan H., Zamparo L., Starczynowski D.T., Leslie C., Papapetrou E.P. Sequential CRISPR gene editing in human iPSCs charts the clonal evolution of myeloid leukemia and identifies early disease targets. Cell Stem Cell. 2021;28:1074–1089.e7. doi: 10.1016/j.stem.2021.01.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yeo N.C., Chavez A., Lance-Byrne A., Chan Y., Menn D., Milanova D., Kuo C.-C., Guo X., Sharma S., Tung A., et al. An enhanced CRISPR repressor for targeted mammalian gene regulation. Nat. Methods. 2018;15:611–616. doi: 10.1038/s41592-018-0048-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yiangou L., Grandy R.A., Morell C.M., Tomaz R.A., Osnato A., Kadiwala J., Muraro D., Garcia-Bernardo J., Nakanoh S., Bernard W.G., et al. Method to synchronize cell cycle of human pluripotent stem cells without affecting their fundamental characteristics. Stem Cell Rep. 2019;12:165–179. doi: 10.1016/j.stemcr.2018.11.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data and analyses for this study are available from the following git hub repository: https://github.com/sunayusluer/CRISPRi-Analysis.