Efficient chromosomal gene modification with CRISPR/cas9 and PCR-based homologous recombination donors in cultured Drosophila cells

Abstract

The ability to edit the genome is essential for many state-of-the-art experimental paradigms. Since DNA breaks stimulate repair, they can be exploited to target site-specific integration. The clustered, regularly interspaced, short palindromic repeats (CRISPR)/cas9 system from Streptococcus pyogenes has been harnessed into an efficient and programmable nuclease for eukaryotic cells. We thus combined DNA cleavage by cas9, the generation of homologous recombination donors by polymerase chain reaction (PCR) and transient depletion of the non-homologous end joining factor lig4. Using cultured Drosophila melanogaster S2-cells and the phosphoglycerate kinase gene as a model, we reached targeted integration frequencies of up to 50% in drug-selected cell populations. Homology arms as short as 29 nt appended to the PCR primer resulted in detectable integration, slightly longer extensions are beneficial. We confirmed established rules for S. pyogenes cas9 sgRNA design and demonstrate that the complementarity region allows length variation and 5'-extensions. This enables generation of U6-promoter fusion templates by overlap-extension PCR with a standardized protocol. We present a series of PCR template vectors for C-terminal protein tagging and clonal Drosophila S2 cell lines with stable expression of a myc-tagged cas9 protein. The system can be used for epitope tagging or reporter gene knock-ins in an experimental setup that can in principle be fully automated.

Figure 1.

I-Sce I and CRISPR-mediated cleavage of chromosomal DNA occur with comparable efficiency. (A) Schematic representation of the double-GFP reporter construct. Note that the first copy of GFP is complete and therefore functional, and the second copy only lacks the initiating methionine (symbolized by the lower case g). The non-rearranged reporter leads to low GFP fluorescence in clonally selected, stable cell lines. Upon cleavage at the intervening I-SceI recognition sequence, repair may occur either via homologous recombination (HR, left side) or non-homologous end joining (NHEJ, right side). While HR rearranges the locus and results in high GFP expression, NHEJ leads to short deletions but retains the low level of GFP fluorescence. (B) Depletion of lig4 (an essential NHEJ factor) and exo1 (an essential HR factor) followed by transfection of an I-SceI nuclease expression vector demonstrates the preferred outcome of reporter fluorescence when either pathway is suppressed. Knock-down of Renilla luciferase served as a control. (C) In clonally selected stable cell lines that express a myc-tagged version of the cas9 nuclease protein in addition to the reporter construct, transfection of an in vitro transcribed, cognate CRISPR RNA (see Figure 2A for sequence details, I-Sce17 refers to the 17-nt-long CRISPR targeting region) leads to a comparable extent of DNA cleavage as the I-SceI nuclease itself (as judged by the proportion of GFP^high cells). A control RNA (CRISPR target sequence 5′- GCGGTGGACCAGCTGCAGC-3′) that does not target cas9 to our reporter did not lead to detectable cleavage and rearrangement.

Figure 2.

Quantitative assessment of cas9-mediated cleavage in living cells using various CRISPR RNA guides. (A) Sequence detail of the I-SceI recognition site region in our reporter construct. The length variant CRISPR RNAs with an NGG trinucleotide protospacer-associated motif (PAM) are shown in blue below the reporter sequence and the CRISPR RNA for testing cleavage efficiency at the 5'-NAG-3' trinucleotide PAM is shown in red above the reporter sequence. (B) Quantitative analysis of CRISPR cleavage activity according to targeting sequence length using our HR reporter system. The proportion of GFP^high cells was determined via two-dimensional analysis of fluorescence-activated cell sorting data (side scatter vs. GFP fluorescence) as this enables more reliable separation of the two populations. Two independent cell clones expressing myc-cas9 and the double-GFP reporter were analyzed; the data are presented as the mean ±SD of three independent biological replicates. (C) Quantitative analysis of CRISPR cleavage activity with an NAG trinucleotide PAM. The experiment was performed essentially as described in B, the I-Sce17 NGG CRISPR serves as comparison. Cleavage activity towards an NAG PAM is detectable but occurs with clearly lower efficiency. Two cell clones were tested in two independent biological replicates each. (D) Extending the CRISPR RNA at its 3′-end does not impair cleavage activity. The experiment was performed and analyzed as in B but a CRISPR RNA with a 3′-extension harboring 18 nt of sequence homology downstream of the CRISPR target site was employed. This extension did not impair cleavage activity (compare with B), but also did not rescue the apparent defect of the I-Sce13GG RNA construct either.

Figure 3.

Using CRISPR-mediated induction of DNA double-strand breaks for genome editing via HR. (A) Schematic representation of the targeting construct design and transfection; the length of the homology arms appended via the PCR primers was 80 nt upstream and 60 nt downstream. Further details, including targeting design, are provided in the supplementary information. (B) Assessment of targeting efficiency using the integration of a GFP moiety at the C-terminus of PGK. The three panels represent flow cytometry analysis 5 days after the transfection (top) and after one and two weeks of blasticidin selection (middle and bottom, respectively). The quantity of PCR product transfected (indicated below the bottom panel) refers to the amount applied to one well in a 96-well plate, the amount of sgRNA employed was 100 ng per well. We compared two stable cell clones expressing myc-cas9 that were independently selected (neomycin and hygromycin resistance) either as untreated cells (red bars) or as cells where the NHEJ factor lig4 had been depleted before transfection (two sequential dsRNA treatments, blue bars). The slightly higher proportion of cells detected as GFP positive after the first split in blasticidin-containing medium may have been caused by a not yet complete antibiotic selection process; non-integrating but transfected cells will be transiently resistant to blasticidin. Since the GFP fluorescence is low (endogenous levels), it is challenging to completely discriminate dying cells from weak GFP positive cells. Note that in the control transfection without any targeting PCR product, no cells could be measured upon selection (nd = none detected). (C) An independent experiment with the neomycin selected myc-cas9 cell clone yielded comparable results. For this experiment, only one dsRNA treatment to deplete lig4 was performed; this may explain the slightly lower efficiency compared with B.

Figure 4.

Molecular characterization of epitope tags introduced at the PGK locus. (A) PCR analysis to verify integration at the PGK locus. The primer locations and expected product sizes are indicated on the left, the specific upstream and downstream primers were chosen outside of the homology-region used for targeting. The PCR products were sequenced to verify their identity. (B) FLP-mediated excision of the resistance cassette. The primer locations and expected product sizes before and after excision are indicated on the left. FLP-mediated excision was further verified by sequencing the PCR products obtained for all three tags after introduction of FLP recombinase. (C) Western blot to verify that a defined protein of the expected size is tagged using our protocol. The blots were probed with monoclonal antibodies against GFP (clone B-2 obtained from Santa Cruz), the FLAG-tag (clone M2 obtained from Sigma) and the Strep-tag (Strep-Mab HRP, obtained from IBA in Göttingen/Germany). Detection of tubulin served as a loading control (below, clone E7 obtained from the Developmental Studies Hybridoma Bank). For each tag, two parallel technical replicates were analyzed ( = parallel transfection and selection, referred to as A and B).

Figure 5.

Short homology arms can direct site-specific integration and increased targeting efficiency with U6-promoter based sgRNA expression. (A) We performed an analogous experiment to Figure 3B using PCR primers with only 29 nt of sequence homology at either end. Although the efficiency is clearly lower than what was observed with longer homology, the short arms are able to direct site-specific integration of the PCR product. (B) Transfection of U6-promoter sgRNA template fusions can further boost the recovery of tag-expressing cells. The amount of sgRNA (obtained by in vitro transcription) or U6-promoter-sgRNA template (obtained by overlap extension PCR) is indicated below the diagram. Note that for the in vitro transcribed sgRNA, the maximal amount used (40 ng) is less than in Figure 3 (100 ng). Blue bars depict the results from transient cas9 expression, while red bars depict the results obtained with a stable cas9 expressing clone (hygromycin-resistance). The bars represent the mean of two fully independent biological replicates, error bars indicate the range of the individual values.

Figure 6.

Fluorescence microscopy of Tub56D-GFP and PGK-GFP. (A) Tub56D-GFP tagging reveals cytoskeletal structures in interphase cells (top row) and the spindle in mitotic cells (bottom row). (B) The GFP-tagged PGK protein is localized to both nucleus and cytoplasm. Live cells were stained with Hoechst 33342, a cell-permeable DNA stain with DAPI-like fluorescence, and imaged on a Leica SP2 confocal microscope. We acquired z-stacks covering the entire thickness of the cells, then calculated the corresponding 2D average projections that are displayed in this figure with the Leica software accompanying the microscope. These high magnification images are representative of the GFP-expressing cells in our cultures. In particular, we did not observe any cells with tubulin-like GFP distribution in the PGK-GFP cultures and vice versa.