A Medium-Throughput Single Cell CRISPR-Cas9 Assay to Assess Gene Essentiality
- 2.6k Downloads
Target selection for oncology is a crucial step in the successful development of therapeutics. Clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9 editing of specific loci offers an alternative method to RNA interference and small molecule inhibitors for determining whether a cell line is dependent on a specific gene product for proliferation or survival. In our initial studies using CRISPR-Cas9 to verify the dependence on EZH2 activity for proliferation of a SMARCB1/SNF5/INI1 mutant malignant rhabdoid tumor (MRT) cell line, we noted that the initial reduction in proliferation was lost over time. We hypothesized that in the few cells that retain proliferative capacity, at least one allele of EZH2 remains functional. To verify this, we developed an assay to analyze 10s-100s of clonal cell populations for target gene disruption using restriction digest and fluorescent fragment length analyses.
Our results clearly show that in cell lines in which EZH2 is essential for proliferation, at least one potentially functional allele of EZH2 is retained in the clones that survive.
This assay clearly indicates whether or not a specific gene is essential for survival and/or proliferation in a given cell line. Such data can aid the development of more robust therapeutics by increasing confidence in target selection.
KeywordsCRISPR Target validation EZH2 Target identification Epigenetics
Acute myeloid leukemia
Clustered regularly interspaced short palindromic repeats
Enhancer of zeste 2
Malignant rhabdoid tumor
Polycomb repressive complex 2
short hairpin RNA
Despite large scale genomic sequencing efforts [1, 2], translation of the resulting data into successful treatment in the clinic has been difficult. Furthermore, many targets reported in the literature to be drivers of cancer have not been confirmed on further analysis [3, 4]. The use of RNA interference (RNAi) for target validation allows the effect of gene knockdown on proliferation to be rapidly assessed ; however, these reagents might not always lead to sufficient mRNA knockdown to evoke a phenotypic effect. For example, an RNAi screen to identify epigenetic dependencies found SMARCA2 as a potential target in SMARCA4-mutant cell lines . Yet the screen failed to identify EZH2 in a SMARCB1 mutant MRT cell line, G-401 , despite the fact that small-molecule-mediated inhibition of EZH2 in this cell line can induce cell death [6, 7]. Similarly, two independent pooled shRNA screens in primary cells from mouse models of MLL–AF9 acute myeloid leukemia (AML) did not identify DOT1L as an essential enzyme for these tumors [8, 9], even though Dot1l knockout mouse models [10, 11] and a small molecule DOT1L inhibitor  have indicated a requirement for DOT1L in MLL-driven AML. These data imply that for certain classes of targets, nearly complete inhibition of enzyme function is required to observe meaningful biological phenotypes.
The CRISPR (clustered regularly interspaced short palindromic repeats)–Cas9 system allows for a rapid, cost-effective and precise knockout of target genes through the introduction of insertions/deletions (indels) [13, 14, 15, 16], and should therefore provide the necessary tools to observe phenotypes that require complete loss of protein function. To test this hypothesis, we carried out a proof-of-concept study using EZH2 as a target for CRISPR–Cas9 mediated knockout in cell lines. EZH2 (Enhancer of Zeste 2) is the catalytic subunit of the polycomb repressive complex 2 (PRC2) and has histone methyltransferase activity that catalyzes the mono-, di- and tri-methylation of lysine 27 on histone 3 (H3K27) [17, 18]. Work using RNAi knockdown and small molecule inhibitors of EZH2 has shown that EZH2 is required for the proliferation of EZH2 mutant lymphoma cells, SMARCB1-deficient MRT cells and cells from other cancer types [6, 7, 19, 20, 21, 22, 23].
As expected, the disruption of EZH2 by CRISPR–Cas9 at early time points led to a decrease in the proliferation rate of an EZH2-dependent cell line, but not an EZH2-independent cell line. However, after 2 weeks the remaining cells in the EZH2-dependent cell line expressed near wild-type levels of EZH2 and showed comparable proliferation rates to a control cell line, whereas the EZH2-independent cell line maintained low levels of EZH2. This suggests that in the EZH2-dependent cell line, the cells in which CRISPR–Cas9 has successfully targeted EZH2 rapidly arrest and/or are lost from the cell population, and that the cells that grow out after two weeks are those in which EZH2 has not been effectively targeted and thus retain at least one functional copy of EZH2. If correct, this hypothesis indicates that carrying out CRISPR–Cas9 studies for individual target genes without analyzing single cell clones could produce misleading results. However, single cell cloning can be time-consuming and a large number of clonal cell populations need to be analyzed to derive accurate conclusions owing to the heterogeneous genotypes that could arise. Therefore, we have developed an assay that allows for the rapid measurement of gene knockout rates in 10’s–100’s of single cell clones. This assay uses both restriction digest analysis and fluorescent undigested fragment length analysis to assess the disruption status of each clone, with the assumption that in a cell line for which a gene is required, very few (if any) clones will emerge in which the gene of interest is homozygously knocked out. Using this protocol we were able to quickly and accurately assess EZH2 essentiality in four cell lines, proving that the assay is an accurate, quick and robust measure of gene essentiality. Importantly, this approach should be applicable to any target gene of interest that is expected to affect cell proliferation or survival, thus providing a crucial insight into gene essentiality and target identification.
The Anti-Proliferative Effect of EZH2 Knockout does not Persist in an EZH2-Dependent Cell Population
Given the dependence of G-401 cells on EZH2 activity, we investigated how cells behaved at later time points after EZH2 knockout. Surprisingly, EZH2 expression had already been restored to near control levels in the G-401 but not RD cells by 15 or 21 days post-infection (Fig. 1e), and the proliferation rate in the G-401 cells was restored to approximately control levels (Fig. 1f and g). As knockout results in a permanent and heritable loss of EZH2, expression cannot be recovered post knockout. Thus, these data suggest that the cells remaining at later time points likely retain at least one functional copy of EZH2. If this hypothesis is correct, the surviving G-401 cells will either retain a wild-type allele of EZH2 or will have in-frame indels that do not affect EZH2 protein function. The DNA breaks made by Cas9 are repaired by the imprecise non-homologous end joining DNA repair pathway and some of these mutations could be in-frame indels that retain function, as has been seen for other genes [14, 25, 26, 27]. These results illustrate the potency of CRISPR–Cas9 for the generation of gene knockouts and also uncover some challenges with using CRISPR–Cas9 to assess the consequences of the deletion of an essential gene for target identification and validation.
Assay Work Flow Summary for Single Cell CRISPR–Cas9 Essentiality Assay
Based on the above data, we set out to design and test an assay that would more rapidly and accurately allow the use of CRISPR–Cas9 reagents to test the essentiality of a single gene. After CRISPR–Cas9 targeting, our assay is designed to examine 1) whether the cells that continue to proliferate in a target-dependent cell line are those that retain at least one in-frame allele (either edited or unedited) and 2) whether cells that continue to proliferate in target-independent cell line have alleles that are mostly edited and have out-of-frame alterations. A common technique to study gene knockout phenotypes is to examine single cell clones and we hypothesized that for a cell line which is dependent on a certain gene, it should be difficult, if not impossible, to generate clonal homozygous knockouts for the gene of interest. Conversely, in cell lines for which a large number of homozygous knockout single cell clones are recovered, it would be unlikely that the growth of that cell line is dependent on the gene of interest. Therefore, we developed a protocol that allowed us to examine gene knockout rates in 10’s to 100’s of single cell clones in a medium-throughput manner. This protocol also assesses whether in-frame or out-of-frame alleles for the gene of interest are produced. Thus, this assay allows for the rapid assessment of gene essentiality.
Proof of Concept for Medium-Throughput Assay with EZH2
Four cell lines were used as a proof-of-concept for this assay: EZH2 inhibitor insensitive RD cells (SMARCB1 wild-type), and the SMARCB1-negative cell lines G-401, G-402 and KYM-1 which are dependent on EZH2 catalytic activity for proliferation . Having verified that these cell lines could generate colonies from single cells at a sufficient rate (Additional file 1: Table S1), each line was infected with an EZH2 sgRNA and Cas9; this sgRNA substantially reduces EZH2 expression and function (Fig. 1a; EZH2 sgRNA #3) with limited predicted off-target cut sites (data not shown) and contains a restriction digest site that spans the Cas9 cut site (Additional file 1: Figure S2). Cells were grown in the presence of puromycin for 17 days (by which time we anticipate that gene editing would have proceeded to completion at the EZH2 locus), followed by single cell dilution.
Although restriction digest analysis is an efficient technique to analyze editing frequency, the restriction digest analysis does not offer complete information about the nature of the indels present at the EZH2 locus. For example, in-frame indels or single base alterations could appear as disruptions by restriction digest, but might not affect protein function. To examine this further, we used fluorescent fragment length analysis on a subset of the clones that had been analyzed by restriction digest. In the RD cells, all of the clones had at least one of the three EZH2 alleles edited (a fragment length smaller or greater than 394 bp), and in over half of the clones, all three alleles had been edited resulting in out-of-frame indels (Fig. 3a and Additional file 1: Figure S4). Conversely, in the EZH2-dependent cell lines, almost all clones had at least one fragment length that was 394 bp, indicating the presence of un-edited or in-frame alterations (Fig. 3b, c and d and Additional file 1: Figure S4). The rare clones in the EZH2-dependent cells that seemed to have no PCR products of wild-type length were those in which the fragment lengths scored just below 394 bp or just above 394 bp (Additional file 1: Figure S4), which could be caused by fragment length analysis assay imprecision.
Sequencing allows further refinement of zygosity
Thus, the combined analyses of the restriction digest data and the fragment length data help to clearly define the zygosity of the clonal cells, allowing for confirmation of EZH2 dependent and non-dependent cell lines. Sequencing analysis can also be used to indicate the presence or absence of an in-frame allele.
CRISPR–Cas9 editing of specific loci offers an alternative method to RNAi and small molecule inhibitors for determining whether or not a cell line is dependent on a specific gene product for proliferation and/or survival. Our initial data indicated that knockout of EZH2 using CRIPSR–Cas9 in a cell line that is dependent on EZH2 function for proliferation resulted in an expected proliferation defect. However, growth of this cell population over a prolonged period indicated that this effect was transient, most likely owing to the outgrowth of cells in which EZH2 function remains intact. In order to test our hypothesis that the cells that grow out after 20 days are those in which EZH2 has either not been targeted, or in which the indels lead to in-frame alterations, we developed a medium throughput assay to analyze 10s–100s of clones for gene essentiality.
The results from our gene essentiality assay clearly show that in cell lines that are EZH2-independent, proliferation rates are unaltered, and the majority of clones have EZH2 alleles that have been edited such that EZH2 function is likely lost. Conversely, in all three dependent cell lines, fewer clones survive the gene knockout process. Of those that do, retention of an in-frame or wild-type copy of one allele is observed in every clone. Thus, the combined analyses of the restriction digest data and the fragment length data help to clearly define the zygosity of the clonal cells, allowing for confirmation of EZH2-dependent and non-dependent cell lines. Sequencing analysis can also be used to indicate the presence or absence of an in-frame allele.
Although this assay can be used to help identify genes that are essential for the growth of specific cancer cell lines, it can also be used to determine whether a gene of interest is likely to be a poor drug target owing to its essentiality in many different cell lines (i.e. broad essentiality would be indicative of potential safety issues in the clinic). Therefore, our gene essentiality assay is applicable to diseases other than cancer. There are specific considerations that need to be addressed when using this assay. First, the sgRNAs chosen for this assay need to be assessed before-hand, specifically showing that they lead to a substantial decrease in target expression and that there are no proliferation effects that are due to off-target effects. In our work with EZH2 we were able to choose a sgRNA that was validated in a cell line in which EZH2 is not required for proliferation. This might be challenging to address for genes of interest in which no target-independent cell line has been identified. There are several algorithms that allow potential off-target binding sites to be assessed for each potential sgRNA and as CRIPSR–Cas9 screens become more widely used, the data generated from these screens should prove useful for the identification of guides that do not have off-target proliferation effects. Second, some genes might not be required for cell line proliferation on tissue culture plastic, but are essential for growth in soft agar or other matrix conditions [28, 29, 30]. Such an effect would be missed in our current assay, but could be easily addressed by including single cell assays in alternative tissue culture conditions. Third, as it is currently configured, this assay cannot be used in cell lines that do not grow or survive as single cells, thus probably limiting its use for primary cells and patient samples. Finally, although this assay can identify indels induced by Cas9 editing, it does not test for enzymatic function of these mutations.
These findings highlight an important caveat to the use of CRISPR–Cas9 for target identification and validation, and present a robust assay to overcome these challenges. We have developed a novel medium-throughput assay to analyze gene essentiality in cell lines that are amenable to single cell dilution cloning. This method should aid both the understanding of whether the protein from a target gene of interest needs to be completely absent for a phenotypic effect or whether reduced expression is sufficient to induce the phenotype of interest. Such data will support the development of more robust cancer therapeutics.
Vector Construction and Virus Production
sgRNA pRSGC1-U6-sg-CMV-Cas9-2A-Puro plasmid and lentivirus for Figs. 1 and 2 were purchased from Cellecta. The sequences for the sgRNA’s are as follows: EZH2 sgRNA #1 ACCGGGGAAGCAGGGACTGAAACG; EZH2 sgRNA #2 ACCGGGTCCCAATTAACCTAGCAA; EZH2 sgRNA #3 ACCGGTGGTGGATGCAACCCGCAA. For additional studies, sgRNA #3 sequence was cloned into pLentiCrispr v2 at DNA 2.0 . HEK293T cells were transfected with the pLentiCrispr-sgEZH2 construct and Virapower packaging system using Lipofectamine 3000 (Life Technologies). Media was replaced with DMEM, 10 % FBS, 1 % BSA after 6 h and harvested 65 h later. Viral particles were concentrated using Lenti-X concentrator (Clontech).
Cells were lysed in RIPA buffer containing 10 % SDS and cOmplete Protease Inhibitor Cocktail Tablets (Roche). Lysates were spun at 16,000 × g at 4 °C for 30 min and normalized for protein concentration. Lysates were separated using SDS-PAGE, and transfer/blotting was performed as described previously . Blots were imaged by fluorescent imaging using the Odyssey infrared imaging system (LI-COR Biosciences; Millennium Science, Surrey Hills, Australia). The following antibodies were used: EZH2 (Cell Signaling #3147), H3K27Me3 (Cell Signaling #9733) and total H3 (Cell Signaling #3683).
Incucyte Proliferation Assays
Cells were plated in triplicate in 96 well plates and confluency measurements were taken every 2 h using an Incucyte Kinetic Imaging System (Essen BioScience).
Functional Titration and Spinfection of Cell Lines
Cells were spinfected by placing one million cells per well in four wells of a 12 well plate in media containing 8 μg/ml polybrene. Multiple concentrations of virus (0–40 μl) were added and the plate was centrifuged at 568 xg for 2 h at 37 °C. Virus containing media was then removed and replaced with 1 ml of media without polybrene prior to incubation at 37 °C. The following day, cells were trypsinized and the cell suspension was divided into two wells of a 6-well plate and a lethal dose of puromycin (0.4 μg/ml) was added to one well. Seventy-two hours after plating, cells were trypsinized and counted and the percent age of transduced cells was determined. Cells infected with the optimal volume of virus were maintained for 17 days in puromycin prior to plating for single cell colonies.
Single Cell Dilution of Cell Lines
All four cell lines had been previously evaluated for their ability to robustly grow from single cells. On the day of plating, cells were trypsinized, counted, diluted to a density of 0.75 and 1.5 cell per well and plated in five 384-well plates at each density by Multidrop dispenser (Thermo Scientific). Plates were imaged using a Cell Metric™ within 6 h of seeding, at intervals of seven days and at the time of harvest. When colonies had reached sufficient size (at least 25 % well confluence), data from the Cell Metric™ was used to generate a pick list for each cell line. Colonies were trypsinized and consolidated from multiple plates into a single 384-well plate for each cell line using a Hamilton Microlab Star robotic system. Samples were immediately lysed in DirectPCR lysis reagent (Viagen).
Restriction Digest Analysis of EZH2 Targeted Colonies
The region around the Cas9 cut site targeted by the sgRNA was amplified by PCR. Primers used were as follows; forward AGGCAAACCCTGAAGAACTG, reverse ATGGCTCTCTTGGCAAAAAT. PCR products were digested with the restriction enzyme FauI, which directly overlapped the Cas9 cut site for the sgRNA targeting EZH2. Digested PCR products were resolved by agarose gel electrophoresis and assessed for the genotype of disruption of the FauI cut site (heterozygous disrupted, homozygous disrupted or homozygous un-disrupted).
Fragment Length Analysis
The region around the Cas9 cut site targeted by the sgRNA in either EZH2 or the ROSA26 locus was amplified by PCR using primers of the same sequence as for the digest analysis, with the forward primer tagged with FAM. Resultant PCR products were purified using Diffinity RapidTips (Sigma) and diluted 1:50 in H2O. Capillary electrophoresis was performed on the samples (Source Biosciences) using Liz-600 size standards; fragment length was determined using Genemapper5 software (Applied Biosystems).
We would like to thank the employees of Epizyme, Inc. for helpful discussions in regards to this publication. We would also like to thank C. Wiggins for informative discussions regarding the development of the assay.
- 19.Sneeringer CJ, Scott MP, Kuntz KW, Knutson SK, Pollock RM, Richon VM, et al. Coordinated activities of wild-type plus mutant EZH2 drive tumor-associated hypertrimethylation of lysine 27 on histone H3 (H3K27) in human B-cell lymphomas. Proc Natl Acad Sci U S A. 2010;107(49):20980–5.PubMedCentralCrossRefPubMedGoogle Scholar
- 24.Chesworth R, Duncan KW, Kawano S, Keilhack H, Klaus C, Kuntz KW, et al. Aryl-or heteroaryl-substituted benzene compounds; WO2012142504 A1. WO2012142504 A1; 2012.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.