A lentivirus-free inducible CRISPR-Cas9 system for efficient targeting of human genes

Abstract

CRISPR-Cas9 is a cutting-edge tool for modifying genomes. The efficacy with which Cas9 recognizes its target has revolutionized the engineering of knockouts. However this efficacy complicates the knocking out of important genes in cultured cells. Unedited cells holding a survival advantage within an edited population can confound the knockout phenotype. Here we develop a HeLa-based system that overcomes this limitation, incorporating several attractive features. First, we use Flp-recombinase to generate clones stably integrated for Cas9 and guide RNAs, eliminating the possibility of unedited cells. Second, Cas9 can be induced uniformly in the clonal cultures using doxycycline to measure the knockout phenotype. Third, two genes can be simultaneously knocked out using this approach. Finally, by not involving lentiviruses, our method is appealing to a broad research audience. Using this methodology we generated an inducible AGO2-knockout cell line showing normal RNA interference in the absence of doxycycline. Upon induction of Cas9, the AGO2 locus was cleaved, the AGO2 protein was depleted, and RNA interference was compromised. In addition to generating inducible knockouts, our technology can be adapted to improve other applications of Cas9, including transcriptional/epigenetic modulation and visualization of cellular DNA loci.

INTRODUCTION

The CRISPR-Cas (clustered regularly interspaced short palindromic repeats-CRISPR associated) system found in several bacteria and archaea embodies an adaptive immune mechanism that relies on faithful recognition of specific nucleic acid sequences (1–9). Detailed investigation of the mechanism by which the CRISPR-Cas system recognizes and degrades foreign DNA sequences has fueled the emergence of a new era of genome editing (10–15). Several clades of CRISPR-Cas mediated interference systems exist, yet Cas9 from Streptococcus pyogenes (SpCas9) is the most extensively used member of the Cas9 endonuclease family (16–19). Although the natural function of SpCas9 (hereby referred to as Cas9) is to cleave infecting phage DNA, Cas9 and its accessory RNA elements have been engineered to recognize and/or cleave DNA, both in vitro (13,14) and in several eukaryotic model organisms (20). Cas9 is an RNA-guided DNA endonuclease that forms a complex with a pair of RNA molecules: a guide or CRISPR RNA (crRNA) and an accessory trans-activating CRISPR RNA (tracrRNA). This ribonucleoprotein complex binds to the genomic target via Watson-Crick base pairing through information provided by the guide RNA, and Cas9 cleaves the double-stranded (ds) DNA target. For genome-editing applications, the crRNA and the tracrRNA can be fused to create a single RNA molecule termed as sgRNA (13). The superior ability of Cas9 to recognize its DNA target, in even the most complex of eukaryotic genomes, is what qualifies Cas9 as a powerful tool for editing genomes. In addition to representing an excellent advancement in biotechnology, CRISPR-Cas9 also holds immense promise for the cure of genetically defined diseases that remain intransigent to other forms of therapy (21,22).

Although the potential of CRISPR-Cas9 in genome editing is clear, there are several other important applications of this technology. Many such applications utilize a version of Cas9 that is catalytically dead (dCas9), but fully capable of binding the DNA target in an sgRNA-dependent manner (23). For example, dCas9 can be directed to promoters for regulating gene expression (24–28). dCas9 can also be tethered to chromatin-modifying enzymes to affect site-specific epigenetic changes (29). Furthermore, dCas9 fused to fluorophores such as GFP may be used to directly visualize specific DNA sequences in living (and fixed) cells (30); a technique which previously required cell fixation-based fluorescence in situ hybridization (FISH) approaches.

Establishment of gene knockouts depends on error-prone non-homologous end joining (NHEJ) of ds DNA breaks created by Cas9. Given that the major determinant of knockout efficiency is activity of Cas9, it is not surprising that this enzyme has been successful in knocking out genes in various biological contexts (31). Although the ability of CRISPR-Cas9 technology to efficiently knockout genes is an extremely attractive characteristic, it also poses disadvantages compared to existing RNA-knockdown technologies. For example, attempts to knock out an essential gene in cultured human cells will result in selection for cells/clones that are unedited (or edited but still preserve gene function). Therefore, the measured cellular phenotype/s of the surviving cells will not be representative of a true gene knockout. Inducing Cas9 uniformly in all cultured cells can circumvent this problem and allow for the detection of an immediate knockout phenotype. Indeed, there are methods to induce Cas9 both in mouse tissues (32) and human induced pluripotent (iPS) cells (33). However these methods require injection/transfection/transduction of guide RNAs into the cell population, leading to the caveat mentioned above. Specifically, in the case of Cas9 targeting essential genes, cells that do not receive guide RNAs and thus remain unedited have a potential growth advantage over edited cells. Finally, most of the published methods involve lentiviral-based approaches for delivering Cas9 and/or the guide RNAs (34,35). Although advantageous in several respects, lentiviral approaches require stricter biosafety considerations that pose an additional obstacle for laboratories that are not equipped/approved to conduct such experiments.

Given that the HeLa cell line remains the most popular experimental tool for studying human gene function in cell culture (36), we set out to develop a robust methodology to apply the CRISPR-Cas9 system in this cell line. We developed a system with the ability to: (i) generate clones integrated stably with single copies of Cas9 and guide RNA genes to eliminate the possibility of unedited cells; (ii) prevent Cas9 induction and allow for the propagation/storage of clones until the time of the knockout experiment; (iii) knockout multiple genes simultaneously; (iv) provide a simple, transient transfection-based, lentivirus-free protocol; and (v) provide an economical method for gene disruption obviating the need for repeated use of siRNA/guide RNA or transfection reagents. By using a combination of a HeLa-based clonal cell line [HeLa EM2-11ht (37)] and a new vector that we describe in this study (pG1G2-FLAG-Cas9-F3), we have developed a method that successfully fulfills all these criteria. As proof-of-principle, we generated an inducible knockout of the AGO2 gene in HeLa cells, and successfully shutdown RNA interference as a function of Cas9 induction. Our choice of AGO2 knockout was driven by our inability to generate clones of this knockout using standard CRISPR-Cas9 protocols, possibly because of the importance of RNAi for cell growth and function. We believe that our newly developed system holds promise not only for the gene-editing functions of Cas9, but also for the numerous other applications of Cas9 that have emerged.

MATERIALS AND METHODS

Reagents and kits for molecular biology

Oligonucleotides for PCR priming, Cas9 guide RNA cloning, and Sanger sequencing were purchased from Integrated DNA Technologies. All restriction enzymes were purchased from New England Biolabs (NEB). Purification of plasmid DNA and other cloning intermediates was performed using DNA purification kits from Qiagen. Genomic DNA from cultured human cells was isolated using the GenElute Mammalian Genomic DNA Miniprep kit from Sigma. PCR reactions for cloning purposes were performed with either Pfu Turbo DNA polymerase (Agilent) or Phusion High-Fidelity DNA Polymerase (NEB) using the manufacturers’ protocols. Ligations of DNA vectors with inserts were performed using the Quick Ligation Kit (NEB). Calf intestinal alkaline phosphatase (CIP) for removing the 5’-phosphate of vectors prior to ligation was purchased either from Promega or from NEB. Site-directed mutagenesis was performed using QuikChange II (Agilent). Reagents for CRISPR-Cas9 experiments are described separately below.

Sanger sequencing

The DNA sequences of all inserts as well as plasmid regions involved in site-directed mutagenesis were verified using Sanger sequencing conducted at the University of Michigan DNA Sequencing Core.

Parental plasmids

The bicistronic vector pX330-U6-Chimeric_BB-CBhhSpCas9 for site-specific genome editing in cultured human cells was a kind gift from Dr. Feng Zhang, Broad Institute of MIT and Harvard, McGovern Institute for Brain Research, and Departments of Brain and Cognitive Sciences and Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA (Addgene plasmid # 42230) (31), and was obtained upon signing a material transfer agreement (MTA). The pX333 vector that allows for cloning of two tandem U6 promoter-driven guide RNAs was a kind gift from Dr. Andrea Ventura, Cancer Biology and Genetics Program, Memorial Sloan Kettering Cancer Center, New York, NY (Addgene plasmid # 64073) (38), and was obtained upon signing an MTA. The pBI-F3-miRNA-d1GFP-loxP-F (abbreviated here as pBi-F3) and pBI4-miRNA-d1GFP (abbreviated here as pBi4) plasmids were obtained from TET Systems GmbH & Co. KG, Heidelberg, Germany, upon signing an MTA, and have been described previously (37,39,40). The FLAG-TPP1 plasmid and the plasmid encoding an shRNA that targets the gene coding for human TPP1 have been described previously (39,41).

Cloning of the g1/g2-inducible Cas9 F3 vector

To obtain a FLAG-Cas9 cassette driven by a doxycycline (dox)-inducible promoter, the pBI4-miRNA-d1GFP vector (6.4 kb) was first digested with NotI and XhoI. The fragment containing the dox-inducible promoter was treated with CIP and ligated to the FLAG-Cas9 insert obtained by PCR amplification of the FLAG-Cas9 fragment in the pX330 vector (containing codon-optimized Streptococcus pyogenes Cas9 gene, a FLAG tag, and two nuclear localization sequences) to yield the pFLAG-Cas9-Bi4 plasmid.

To insert the tandem guide RNA expression cassette into the pBi-F3 backbone, the pBI-F3-miRNA-d1GFP-loxP-F vector was digested with BglII and StuI. The ~5.4 kb fragment obtained after gel-purification of the digestion reaction was treated with CIP and ligated to the tandem guide RNA expression cassette insert amplified using PCR of the pX333 vector template with to yield the pG1G2-F3 plasmid.

To sub-clone the FLAG-Cas9 fragment in the pG1G2-F3 backbone, the pG1G2-F3 and pFLAG-Cas9-Bi4 vectors were separately digested with StuI and HpaI. The restriction digestion of pG1G2-F3 yielded a ~1.4 kb fragment and a ~4.9 kb fragment. The ~4.9 kb fragment isolated after gel purification was treated with CIP prior to DNA ligation. The restriction digestion of pFLAG-Cas9-Bi4 yielded a ~4.8 kb fragment and a ~3.8 kb fragment. The ~4.8 kb fragment was gel-purified and ligated to the purified vector fragment to yield the pG1G2-FLAG-Cas9-F3* vector. This vector contains a dox-inducible FLAG-Cas9 gene and the sites for cloning in two guide RNA genes. However, the BbsI recognition site, which is the cloning site for one of the two possible guide RNAs, is repeated at three additional sites in the vector. While one extraneous BbsI site (most proximal to the guide RNA cassette) was removed using site-directed mutagenesis (Agilent), the other unwanted BbsI sites were subsequently and sequentially mutated using two-step overlap extension PCR methodology (42). This yielded the final pG1G2-FLAG-Cas9-F3 vector that contains (i) a dox-inducible FLAG-Cas9 gene; (ii) two tandem U6 promoter-driven cloning sites for guide RNA expression; and (iii) the F/F3 FRT sites that allow for integration into a homologous site in the HeLa EM2-11ht genome.

Cell culture

All experiments in this study involving cultured human cells were performed with HeLa EM2-11ht cells or its clonal derivatives developed for this study. The HeLa EM2-11ht cell line was obtained from TET Systems GmbH & Co. KG, Heidelberg, Germany, upon signing a MTA. Cells were cultured in an incubator maintained at 37 °C and 5% CO₂ in growth medium containing DMEM, 10% FCS, 2 mM L-glutamine, 1 mM sodium pyruvate, 100 units/ml of penicillin, and 100 mg/ml of streptomycin. Induction of dox-inducible genes in this cell line was performed using doxycycline (Sigma) added to the growth media at a final concentration of 200 ng/ml.

Immunoblotting

Standard immunoblot protocols were used with the following primary antibodies at specified dilutions: mouse monoclonal anti-Flag M2-HRP conjugate (Sigma; A8592; 1:20,000), mouse monoclonal anti-β-actin antibody (Sigma; A5441; 1:10,000), and mouse monoclonal anti-AGO2 antibody (11A9) (43), a kind gift from Dr. Gunter Meister, Center for Integrated Protein Science Munich (CIPSM), Germany. Secondary horseradish peroxidase-conjugated goat antibodies against rabbit IgG (Santa Cruz Biotechnology; 1:10,000) and mouse IgG (Santa Cruz Biotechnology; 1:10,000) were used to recognize the primary antibodies and allow for detection of the protein-of-interest using chemiluminescence detection technology (Pierce ECL Western Blotting Substrate; Thermo Scientific). The data were visualized using a gel-documentation system (ChemiDoc MP System; BioRad).

Transient transfections

All transient transfection experiments were performed using Lipofectamine 2000 (Life Technologies) following the manufacturer’s recommendations. For transfections, cells were seeded ~24 hours prior to transfection at 40–50% confluency. Transfections were performed in growth medium devoid of antibiotics and serum. Medium complete with antibiotics and serum was used to replace the transfection medium 5 h post-transfection.

Guide RNA design

The Zhang Lab CRISPR design algorithm was used to design guide RNAs against AGO2 and DICER (http://crispr.mit.edu/). Based on a high CRISPR design tool score and a low probability of off target cleavage events, one or two of the highest scoring guide RNAs were used in the experiments. The selected guide sequence oligonucleotides were annealed to each other, phosphorylated with T4 polynucleotide kinase enzyme (NEB), and cloned into the BbsI and/or BsaI site/s of the pX333 vector or the pG1G2-FLAG-Cas9-F3 vector for guide RNA expression.

Surveyor nuclease assay

Surveyor nuclease experiments to detect genomic editing by Cas9 were performed by adapting protocols published previously (40). For experiments involving transient transfection of Cas9/guide RNA-containing plasmids, ~0.3 × 10⁶ HeLa EM2-11ht cells were transfected in a 12-well growth format with 1 µg of Cas9/guide RNA-containing plasmids using 2.5 µl Lipofectamine 2000 (Life Technologies). After 72 h of transfection, cells were trypsinized and harvested by centrifugation. 150–200 ng of genomic DNA isolated from the harvested cells was used for PCR amplification with Surveyor primers in a 50 µl reaction volume. Upon completion of the PCR reaction, 4 µl of each reaction was loaded on a 1–2% agarose-TAE gel to check the uniformity and efficiency of amplification among different samples. The amplification products were purified using a PCR purification kit and subjected to the Surveyor nuclease assay according to the manufacturer’s protocol (Surveyor Mutation Detection Kit; catalog #706025; Transgenomic). Upon completion of the Surveyor nuclease cleavage reaction, the products were visualized using ethidium bromide-stained, 2% GTG agarose-TAE gels.

Lentivirus-free production of clonal stable cell lines encoding dox-inducible Cas9

HeLa-EM2-11ht cells were transfected in a 6-well format using Lipofectamine 2000 (Life Technologies) with 1 µg of the indicated pG1G2-FLAG-Cas9-F3 and 1 µg of a puromycin resistance-encoding, Flp recombinase-expressing plasmid. ~24 h post-transfection, cells were treated with puromycin (5 µg/ml; Sigma-Aldrich). After ~24 h of selection under puromycin, fresh medium containing ganciclovir (50 µM; Sigma-Aldrich) was added and selection was conducted for ~10 days. 24 individual clones were picked and transferred to individual wells of a 96-well plate. Once confluent the cells in each well were passaged into duplicate 96-well plates. One of the duplicate plates was used to identify positive clones. Positive clones were identified based on a strong signal in a spot-blot based immunoblotting assay against the FLAG tag (present on the Cas9 construct). For this, clones were grown in a 96-well format in the absence of doxycycline. The plate was trypsinized and split into two new 96-well plates (master plate and replica plate). The replica plate clones were treated for 24 h with doxycycline (200 ng/ml in culture medium) and lysed in 100 µl of 5 M urea-containing buffer. 50 µl of the lysate was blotted on a nitrocellulose membrane using a vacuum manifold-based dot blot apparatus. The membrane was stained with Ponceau-S staining solution (Sigma) and imaged to ensure uniform blotting. The blot was then blocked with StartingBlock (PBS) blocking buffer (Thermo Scientific) and incubated with anti-Flag M2-HRP conjugate (Sigma; A8592; 1:20,000). Positive clones marked on the replica plate were used to expand corresponding clones from the master plate. One positive clone for each pG1G2-FLAG-Cas9-F3 construct was expanded until it grew to confluency in 6 cm dishes. Doxycycline (200 ng/ml) was added to the growth medium to induce genome editing uniformly in cells of the clonal culture.

RESULTS

Generation of a Flp recombinase-targetable vector for stable expression of two guide RNAs and a dox-inducible FLAG-Cas9 protein in HeLa cells

To build a system that allowed for virus-free, stably inducible, genome editing in HeLa cells, we first constructed a vector that encodes for the components that are key to attaining these characteristics (see Materials and Methods for detailed information). The pBi-F3 vector contains a bi-directional dox/tet-inducible promoter (Ptet) as well as a wild-type/mutated F/F3 pair of Flp recombinase targeting sites flanking the expression cassette (37). We and others have previously used this bidirectional cassette to express a protein-coding gene and a non-coding RNA (e.g., shRNA/miRNA) gene simultaneously from CMV promoters in a dox-dependent manner (37,39). We set forth to construct a plasmid that encoded both Cas9 and a cloning site for guide RNAs in the pBi-F3 backbone. However, Cas9 guide RNAs have been expressed in human cells driven by a U6 promoter. The U6 promoter allows for transcription by PolIII, which unlike PolII (which recognizes CMV promoters) does not append a polyA tail at the 3’ end of the guide RNA/spacer RNA sequence. To adapt the pBi-F3 vector to co-express two Cas9 guide RNAs, we first amplified a U6-guide cloning RNA cassette using the pX333 plasmid as a template (38) (Fig. 1). Ligation of the dual guide RNA cloning module with the pBi-F3 backbone resulted in an intermediate vector termed here as pG1G2-F3 (Fig. 1). We note that the selection of cloning sites to introduce the guide RNA cassette into the pBi-F3 vector resulted in the removal of one Ptet element. Thus, the guide RNAs cloned into pG1G2-F3 will be expressed constitutively in cultured human cells. This did not detract from the overarching goal of obtaining an inducible Cas9 vector system.

Figure 1.

Cloning strategy for generation of a vector allowing for site-specific virus-free integration into the genome, and stable co-expression of doxycycline-inducible Cas9 and two guide RNAs. The stepwise cloning protocol starting from FLAG-Cas9 (from pX330) and tandem guide RNA gene inserts (from pX333) leading to the engineering of the final pG1G2-FLAG-Cas9-F3 vector is shown. Restriction sites pertinent to the subsequent cloning step are indicated. F3 indicates a mutated FRT site, while F indicates a wild-type FRT site. The F/F3 combination allows for directional integration into the analogous (and unique) F/F3 site in the HeLa EM2-11ht genome. Ptet indicates a promoter that is induced by addition of dox in cells (e.g., HeLa EM2-11ht cells) also expressing the rtTA2S–M2 protein. The U6 promoter allows for PolIII-driven transcription of guide RNA.

The next step in generating our proposed vector was to insert the FLAG-Cas9 gene into the pG1G2-F3 vector downstream of the remaining Ptet element. We first proceeded to amplify the FLAG-Cas9 gene from the pX330 vector (31) and inserted it into the pBi4 backbone using conventional restriction site-based cloning to yield the pFLAG-Cas9-Bi4 plasmid backbone (Fig. 1). Next the FLAG-Cas9 fragment was sub-cloned from the pBi4 backbone into the pG1G2-F3 vector to yield the pG1G2-FLAGCas9-F3* plasmid (Fig. 1). The resulting plasmid encodes most of the desired characteristics of the final vector, but was not compatible with cloning of the second guide RNA (g2) using the BbsI restriction enzyme. This is because pG1G2-FLAG-Cas9-F3* contained three additional BbsI sites elsewhere on the vector backbone. The unwanted BbsI sites were mutated sequentially using mutagenesis strategies (detailed in Materials and Methods) to yield the final pG1G2-FLAG-Cas9-F3 vector (Fig. 1). The final vector was sequenced across the guide RNA cassette and the Cas9 gene to rule out any unwanted mutations that may have been introduced during the process of cloning.

Cas9 in pG1G2-FLAG-Cas9-F3 can be induced to efficiently edit the HeLa genome

To test the functionality of pG1G2-FLAG-Cas9-F3 we constructed three derivatives of the vector, one encoding a guide RNA against human AGO2, another encoding a guide RNA against human DICER, and a third vector encoding one guide RNA for AGO2 and one guide RNA for DICER (Fig. 2A). To test the dox-dependent induction of FLAG-Cas9, we transiently transfected HeLa EM2-11ht cells. We selected these cells for two reasons. First, this clonal HeLa cell line constitutively expresses the reverse tetracycline controlled transactivator gene coding for rtTA2S-M2 (37). This allows us to switch on genes driven by the Ptet promoter in the presence of dox (50–200 ng/ml). Second, this cell line contains a single, genomically annotated, F/F3 FRT locus for integration of the guide RNA/Cas9 cassette from pG1G2-FLAG-Cas9-F3 to facilitate stable cell line generation (37) (described later). Immunoblotting analysis revealed that transient transfection of pAGO-G1-FLAG-Cas9-F3 and pDICER-G1-FLAG-Cas9-F3 resulted in robust expression of FLAG-Cas9 in HeLa EM2-11ht cells, but only in the presence of dox (Fig. 2B). Thus, as previously reported (37), the Ptet promoter is induced strongly (several-fold above background levels) in the presence of dox in cells also expressing rtTA2S-M2.

Figure 2.

pG1G2-FLAG-Cas9-F3 can be adapted to efficiently cleave the endogenous human AGO2 and DICER loci. (A) The vector diagrams of the three pG1G2-FLAG-Cas9-F3 derived plasmids encoding gRNAs against the indicated gene loci are shown. The pAGO-G1-FLAG-Cas9-F3 and pDICER-G1-FLAGCas9-F3 plasmids both contain one empty guide RNA cassette. (B) HeLa EM2-11ht cells were transfected with plasmids encoding guide RNAs (as shown in panel A) targeting the indicated genes. Immunoblotting was performed on lysates of cells 24 h post-transfection using the indicated antibodies. “+ Dox” indicates transfections including 200 ng/ml of dox in the growth medium, while “− Dox” indicates transfections not supplemented with Dox. (C) The Surveyor nuclease assay was performed to detect cleavage of the indicated genomic targets by guide RNAs and Cas9 derived from the two plasmids shown in panel A. The sizes of the PCR amplicons that were subjected to Surveyor nuclease treatment are shown schematically as horizontal bars at the top. For each experiment, the sizes of the expected Surveyor reaction products, which correspond roughly to the distances (in bp) between the Cas9 cut site (shown as an arrowhead on the schematic) and the ends of the PCR amplicon are also indicated. The digested fragments obtained experimentally from Surveyor nuclease are indicated with arrowheads adjacent to the gel, and match closely the predicted sizes. (D) Surveyor assay results for the indicated guide RNAs (AGO2 or DICER) residing on the indicated plasmid backbones (pX333 or Cas9-F3) is shown. Cleavage efficiency of a given guide RNA was similar between the two tested vector backbones.

Next, we asked if the expressed Cas9 protein in combination with the guide RNA was able to cleave the human genome in a sequence-specific manner. To measure genome editing via CRISPR-Cas9 action, we performed the Surveyor nuclease assay that we and others have reported previously (31,40). This assay depends on Surveyor nuclease-catalyzed cleavage of mismatched regions in an otherwise perfect DNA duplex. In human cells, efficient cleavage by Cas9 is generally followed by rejoining of broken ends through NHEJ. This error-prone DNA ligation pathway frequently introduces variable insertions/deletions (indels) at the site of Cas9 cleavage. With such a differentially edited cell population, PCR amplification of the DNA spanning the Cas9 recognition sequence followed by denaturation/renaturation will result in DNA duplexes containing mismatches around the Cas9 cut site. These mismatches, which serve as a scar of Cas9 action, become a substrate for the Surveyor nuclease. Indeed, induction of Cas9 for 72 h post-transfection resulted in robust cleavage of the human AGO2 and DICER loci with guide RNAs targeting these loci (Fig. 2C). No cleavage was detected in the absence of dox, further verifying the inducible nature of the CRISPR-Cas9 knockout using this system.

Next we compared the genome editing capacity of the pG1G2-FLAG-Cas9-F3 vector to that of a well-established Cas9 vector, pX333. For this, we cloned the AGO2 guide RNA into the pX333 vector either individually or in combination with the DICER guide RNA. Cleavage of AGO2 and DICER in the cells expressing the appropriate guide RNAs encoded by the pG1G2-FLAG-Cas9-F3 vector backbone was comparable to that of transfections of pX333-derived plasmids encoding the same guide RNAs. These data suggest that neither our cloning strategy, the nature of the Ptet promoter, nor the characteristics of the HeLa EM2-11ht cell line compromised the efficacy of CRISPR-Cas9 in genome editing to any measurable extent (Fig. 2D). Additionally, the ability of the pAGO-DCR-FLAG-Cas9-F3 plasmid to simultaneously orchestrate AGO2 and DICER cleavage validates the use of this vector for efficiently co-expressing two separate guide RNAs.

Virus-free generation of a cell line stably encoding a guide RNA targeting AGO2 and an inducible Cas9 gene

The final step towards developing a lentivirus/retrovirus-free system to stably integrate guide RNAs and a dox-inducible Cas9 gene involves integrating the guide RNA-Cas9 module into the unique F/F3 FRT locus of the HeLa EM2-11ht cell line. The unique integration site ensures uniform expression of inserted genes in a clonal population, compared to lentiviral methods which integrate inserts randomly across the host genome. As proof-of-principle, we decided to invoke two pG1G2-FLAG-Cas9-F3 constructs (separately), one that encodes a guide RNA targeting AGO2 (to yield AGO2-g1-Cas9-F3 HeLa), and another that is devoid of any guide RNA sequence (to yield Empty-Cas9-F3 HeLa). HeLa EM2-11ht cells were co-transfected with the appropriate FLAG-Cas9-F3-guide RNA construct, and a second plasmid encoding Flp recombinase and a puromycin resistance marker (Fig. 3A). Transfected cells were selected for by including puromycin in the growth medium for ~24 h. Flp recombination-driven integration of the guide RNA-Cas9 locus between the F/F3 sites in the HeLa EM2-11ht genome will result in loss of the thymidine kinase gene that normally separates the F and F3 sites in the parental genome (Fig. 3A). Thymidine kinase expression in the parental cell line renders it sensitive to ganciclovir (gan), allowing for selection of successful recombinant clones by including gan in the growth medium. After ~10 days of gan selection, clones were readily visible under a light microscope. 24 such clones were initially passaged into a 96 well plate, from which duplicate cultures were established. Cells from one of the duplicate cultures (still in 96-well format) were induced briefly (~24 h) with dox and harvested for screening (Fig. 3A). We performed Western spot-blotting for the FLAG epitope (present on the FLAG-Cas9 construct) to screen for successfully recombined clones (Fig. 3A, B). As shown in Fig. 3B, positive clones were easily recognizable due to the substantially higher than background anti-FLAG signal. Only the un-induced replicate cultures were utilized for downstream analysis. We proceeded to characterize one clone each of AGO2-g1-Cas9-F3 HeLa and Empty-Cas9-F3 HeLa.

Figure 3.

Virus-free generation of a cell line stably expressing dox-inducible Cas9 and guide RNAs against human AGO2. (A) A schematic for the engineering of HeLa EM2-11ht clonal cell lines stably integrated with the guide RNA/FLAG-Cas9 cassette using transient Flp recombinase expression. Transfected cells are enriched for using selection with puromycin, resistance to which is provided by the Flp-encoding plasmid. Stably integrated cells are enriched for by treatment with ganciclovir, which causes toxicity in parental cells because of continued thymidine kinase expression. Colonies surviving the selection procedure were expanded and analyzed for dox-dependent FLAG-Cas9 expression using a spot-blot based anti-FLAG immunoblot assay. (B) Anti-FLAG spot-immunoblot of positive clones for Empty-Cas9-F3 HeLa (well # C1 in a 96-well plate) and AGO2-g1-Cas9-F3 HeLa (well # A7 in a 96-well plate) are shown. Dox was added to induce Cas9 expression.

Induction of Cas9 results in cleavage of the genomic target, reduction of target protein levels, and loss of target protein function

AGO2 is a central component of the RNA interference machinery in human cells (44–47). Deletion of the AGO2 gene is therefore expected to diminish siRNA/shRNA-mediated RNA silencing in cultured human cells (Fig. 4D). We proceeded to test this hypothesis using our newly established AGO2-g1 Cas9-F3 HeLa cell line. First, we investigated the ability of dox to switch on FLAG-Cas9 expression in the experimental cells as well as in the “Empty Cas9-F3 HeLa” cells. Indeed, robust Cas9 expression was established within 1 day of induction with dox in both the control and the experimental cell lines (Fig. 4A). Although we did notice a faint Cas9 signal (which was more prominent in the control cell line) at day 0, we believe this reflects the high sensitivity of the FLAG-antibody rather than the Cas9 protein levels necessary for detectible biological function (see below). Given that maximal/stable Cas9 levels are obtained within the first tested time point, our method provides for rapid Cas9 induction.

Figure 4.

Induction of Cas9 in the presence of an AGO2-targeting guide RNA results in robust cleavage of the human AGO2 locus, rapid depletion of AGO2 protein, and loss of RNA interference capacity. (A) Empty-Cas9-F3 HeLa and AGO2-g1-Cas9-F3 HeLa cells were treated with dox (200 ng/ml) and induction of FLAG-Cas9 was monitored as a function of days of induction using anti-FLAG immunoblotting. “0 days in dox” indicates a time-point immediately prior to dox treatment. (B) Surveyor analysis as described in Fig. 2C and D was performed with genomic DNA derived from the indicated stable cell lines after indicated durations of dox treatment. AGO2 cleavage was absent in the Empty vector cell line, while clearly present in AGO2-g1-Cas9-F3 HeLa cells at all time-points measured after dox treatment. Arrowheads indicate the Surveyor cleavage products, which match the predicted sizes shown in the schematic in Fig. 2C. (C) Endogenous AGO2 protein levels were detected using immunoblotting; AGO2 levels sharply declined upon dox addition in AGO2-g1-Cas9-F3 HeLa cells. (D) Schematic of the prediction of consequences of AGO2 knockout on FLAG-TPP1 expression. (E) Results of the sh^TPP1 knockdown experiment in control and AGO2-g1-Cas9-F3 HeLa cells induced with dox. While the shRNA efficiently knocks down TPP1 in the control cells, it is unable to do so in cells AGO2-deficient AGO2-g1-Cas9-F3 cells.

Next, we performed the Surveyor nuclease assay to investigate the ability of the genomically encoded guide RNA and Cas9 protein to edit the AGO2 locus. Consistent with the kinetics of Cas9 induction, maximal AGO2 cleavage was observed at the earliest time-point after induction (Fig. 4B). Although it seemed as though only a small fraction of the amplicon was cleaved by the Surveyor nuclease, we envision the genome editing to be almost (or fully) complete in the cultured population for the following reasons. First, we believe that it is not possible to observe 100% cleavage in the Surveyor nuclease assay (even with the manufacturer’s positive control; data not shown), because the Surveyor nuclease is limiting under these conditions. Second and more importantly, the highest running band in the AGO2-g1-Cas9 HeLa cells runs slightly, but consistently, lower than the corresponding band (representative of Cas9 un-cleaved DNA) in the empty vector control (Fig. 4B). This is consistent with the observation of a greater accumulation of deletions (versus insertions) as a result of NHEJ-based repair (48). The third and most conclusive evidence for efficient knockout of AGO2 in the CRISPR-edited cells comes from immunoblot analysis of endogenous human AGO2 protein. Whereas AGO2 protein levels are unaffected in control cells or in AGO2-g1-Cas9 HeLa cells before dox treatment, AGO2 protein levels rapidly decline upon Cas9 induction (Fig. 4C).

In our previous studies on the telomere protein TPP1, we have described an shRNA construct (sh^TPP1) that efficiently knocks down recombinant FLAG-TPP1 (as well as endogenous TPP1) in HeLa EM2-11ht cells (39). We hypothesized that inducing AGO2 cleavage will diminish RNA interference (RNAi) in these cells, leading to a restoration of TPP1 protein levels despite the presence of its targeting shRNA (Fig. 4D). Fully consistent with this idea, sh^TPP1 efficiently knocked down transiently expressed FLAG-TPP1 in Empty-Cas9-F3 HeLa cells, but was not effective in reducing TPP1 protein levels in AGO2-g1-Cas9-F3 HeLa cells (Fig. 4E). In summary, we have developed a plasmid-based, virus-free system for conducting genome editing in an inducible manner uniformly across cells in a HeLa culture. Using this system, we efficiently induced a human AGO2 knockout. As a result, we were able to assess the immediate biological consequences of AGO2 deletion, namely, the disruption of RNAi.

DISCUSSION

CRISPR-Cas9 technology has revolutionized the field of genome editing. Recent and rapid developments have allowed for significant improvement in precision, efficiency, as well as versatility of this technology. Our study provides a significant improvement for the application of CRISPR-Cas9 technology in cultured human cells. Although technologies exist for inducing Cas9 expression, engineering cell lines to stably integrate Cas9, and utilizing recombinogenic methods to insert Cas9 into the genome, our study is the first (to our knowledge) to integrate all these features in one system for use in cultured human cells. Specifically, we developed a FIp-recombinase targetable vector that contains a dox-inducible Cas9 gene as well as the capacity to express two guide RNAs. Transient transfection of this vector results in efficient Cas9 expression and genome editing, but in a strictly dox-dependent manner. More importantly, integration of this module into the genome of HeLa cells provides the ability to edit genomes in an inducible manner. As proof-of-principle, we knocked out AGO2 in a clonal culture stably co-expressing Cas9 protein and guide RNAs against AGO2 to attenuate RNAi. The lack of lentiviral intervention in our methodology makes it attractive to a wider spectrum of human cell biologists. Our method requires ~one month from start to completion, providing a swift avenue for generating inducible knockout cell lines of potentially any gene in the human genome (Figure 5).

Figure 5.

Timeline for generating stable cell lines expressing dox-inducible Cas9 and guide RNAs.

AGO2 is a gene critical for mammalian development and growth (46,49). Our ability to knock this gene out in cultured human cells highlights an important application of our methodology. Our method involves co-expression of the relevant guide RNAs and Cas9 uniformly in clonal culture, severely reducing the possibility of unedited cells in the population. It can be argued that although our method will uniformly edit all cells, cells where editing did not disrupt gene function substantially (e.g., short in-frame mutations) would hold a survival advantage in culture. However, this concern can be overcome easily using two guide RNAs to excise a large region of the gene-of-interest. Hence, we believe that our method holds the potential to reveal the immediate consequences of knockout of any essential human gene in cell culture.

We envision several other advantages as well as potential applications of our newly developed technololgy. First, the use of stably inducible cell lines in our methodology provides an obvious economical advantage. Once established for any particular gene-of-interest, it eliminates the need for repeated transfections. Second, the number of guide RNAs that can be expressed using our method is not limited to two. The pG1G2-FLAG-Cas9-F3 can be easily adapted to insert multiple copies of the dual-guide RNA cassette to facilitate multiple gene knockouts (e.g., of all/several redundant genes in a pathway) using a single vector. Third, our strategy is not only restricted to generation of gene knockouts, but can also be very useful in generating gene knock-ins. Knock-in experiments involve co-transfection of the guide RNA/Cas9 vector and the DNA repair template, which is often a single-stranded oligodeoxynucleotide (ssODN) containing the intended mutation. Our method eliminates the necessity of transfecting multiple DNA/RNA constructs, and avoids the complications that accompany the co-transfection of nucleic acids of different types (single-stranded vs. double-stranded). Using our method, only the repair ssODN needs to be introduced into the Cas9-guide RNA encoded cells prior to induction of Cas9 expression.

Fourth, our method will allow for a vast improvement in the exploitation of dCas9-GFP as a replacement over conventional fluorescence in situ hybridization techniques. For example, dCas9-GFP has been used to visualize telomeres and other genomic loci in human cells (30). Although these experiments serve as proof-of-principle for application of CRISPR-Cas9 technology, they suffer from potential drawbacks. For example, constitutive localization of dCas9-GFP and guide RNAs at the genomic target might have unanticipated physiological consequences in the cell. At telomeres, such binding could result in unwanted chromosome end-deprotection and/or telomerase misregulation (50). Our method could be easily adapted to attenuate such side effects. A pBi-F3-based dCas9 vector can be stably integrated into the HeLa genome without risk of constitutive expression of the dCas9-GFP protein. A brief pulse of dCas9-GFP expression induced with dox can be used immediately prior to imaging of the telomeres (or other DNA loci), minimizing the risk of unwanted affects from dCas9-GFP association with its genomic target. Similarly, our methodology could improve the application of dCas9 as a modulator of transcription and epigenetic modulation. The inducible (and stable) nature of our Cas9 expression platform will allow for the generation of cell lines where a particular endogenous gene-of-interest may be activated/repressed or epigenetically modified with great temporal control. In summary, our newly developed method provides several improvements over existing CRISPR-Cas9 technology, and holds great promise for the study of cellular function in the context of cultured human cells.